diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,62394 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9997114135924199,
+  "eval_steps": 500,
+  "global_step": 10394,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-08,
+      "loss": 6.7676,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 5.8442,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 5.9004,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 5.915,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128205e-07,
+      "loss": 5.9912,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 6.9463,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4871794871794876e-07,
+      "loss": 6.5898,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 5.8799,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 6.2959,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 6.1855,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.051282051282052e-07,
+      "loss": 6.4883,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 6.0229,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 6.2061,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 5.8955,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 5.9209,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 6.625,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0897435897435899e-06,
+      "loss": 5.8594,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 5.8696,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.217948717948718e-06,
+      "loss": 5.4653,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 5.7222,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 5.6094,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 5.7588,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4743589743589745e-06,
+      "loss": 5.251,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 4.5693,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.602564102564103e-06,
+      "loss": 4.4526,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 4.8472,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 4.2959,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 4.2061,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8589743589743592e-06,
+      "loss": 4.1494,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 4.3286,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.987179487179487e-06,
+      "loss": 3.918,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 3.9917,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 2.9106,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 2.9355,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.243589743589744e-06,
+      "loss": 3.1616,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 3.0718,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.371794871794872e-06,
+      "loss": 2.8281,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 2.6748,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 2.6382,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 2.7087,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6282051282051286e-06,
+      "loss": 2.6616,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 2.4224,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.756410256410257e-06,
+      "loss": 2.3958,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 2.5833,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 2.7825,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 2.302,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.012820512820513e-06,
+      "loss": 2.1934,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 2.1887,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.141025641025641e-06,
+      "loss": 2.1543,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 2.0654,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 2.281,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 2.0171,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.397435897435898e-06,
+      "loss": 2.0513,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 2.0948,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5256410256410263e-06,
+      "loss": 1.9324,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 2.021,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 2.0369,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.8584,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.782051282051282e-06,
+      "loss": 1.9773,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.8682,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.910256410256411e-06,
+      "loss": 1.9656,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 2.0039,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 1.8047,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.8575,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 2.0957,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.8423,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.294871794871795e-06,
+      "loss": 1.9116,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 2.0596,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 1.8472,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 1.7417,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.551282051282052e-06,
+      "loss": 1.7747,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.751,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6794871794871795e-06,
+      "loss": 1.8295,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 2.0139,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 1.7427,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.699,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.935897435897436e-06,
+      "loss": 1.7092,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 1.6758,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.064102564102565e-06,
+      "loss": 1.7426,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.7329,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 1.668,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 1.7312,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.320512820512821e-06,
+      "loss": 1.6456,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.6367,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.448717948717949e-06,
+      "loss": 1.6819,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 1.686,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 1.671,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.7922,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.705128205128206e-06,
+      "loss": 1.6372,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.9099,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.6919,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 1.6887,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 1.6748,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 1.6534,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.08974358974359e-06,
+      "loss": 1.6146,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.478,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.217948717948718e-06,
+      "loss": 1.5696,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.5621,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 1.5377,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 1.7224,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.474358974358975e-06,
+      "loss": 1.5959,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.4772,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.602564102564103e-06,
+      "loss": 1.5705,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.5111,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 1.6255,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 1.4895,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.858974358974359e-06,
+      "loss": 1.6819,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.5345,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.9871794871794876e-06,
+      "loss": 1.4559,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 1.486,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 1.543,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 1.594,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.243589743589744e-06,
+      "loss": 1.4921,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.6106,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.371794871794873e-06,
+      "loss": 1.5999,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 1.4569,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.5322,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 1.5454,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6282051282051286e-06,
+      "loss": 1.5018,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.5475,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.756410256410258e-06,
+      "loss": 1.5516,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 1.4485,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 1.4652,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 1.4531,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.012820512820515e-06,
+      "loss": 1.4291,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.5999,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.141025641025641e-06,
+      "loss": 1.5034,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.4772,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 1.4475,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.5463,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.397435897435898e-06,
+      "loss": 1.3824,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.6248,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.525641025641026e-06,
+      "loss": 1.5101,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 1.3073,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 1.5664,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.4857,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.782051282051283e-06,
+      "loss": 1.4459,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.4409,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.910256410256411e-06,
+      "loss": 1.4591,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.4421,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 1.5791,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 1.4506,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.5016,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.4176,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.294871794871796e-06,
+      "loss": 1.502,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 1.4973,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 1.3982,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 1.5336,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.551282051282053e-06,
+      "loss": 1.4464,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 1.4145,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.67948717948718e-06,
+      "loss": 1.3666,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 1.5278,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 1.3563,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 1.4304,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.935897435897437e-06,
+      "loss": 1.4421,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4474,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0064102564102565e-05,
+      "loss": 1.4631,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 1.3688,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0192307692307692e-05,
+      "loss": 1.4746,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.3652,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0320512820512822e-05,
+      "loss": 1.4028,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 1.3888,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.044871794871795e-05,
+      "loss": 1.4666,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 1.3813,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0576923076923078e-05,
+      "loss": 1.3105,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0641025641025643e-05,
+      "loss": 1.4078,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0705128205128205e-05,
+      "loss": 1.3583,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.3738,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 1.3923,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 1.3773,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0961538461538464e-05,
+      "loss": 1.446,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 1.3581,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1089743589743592e-05,
+      "loss": 1.3259,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 1.4552,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1217948717948719e-05,
+      "loss": 1.3886,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 1.387,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1346153846153847e-05,
+      "loss": 1.2837,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 1.3651,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1474358974358974e-05,
+      "loss": 1.4528,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 1.3971,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1602564102564104e-05,
+      "loss": 1.324,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 1.5238,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1730769230769232e-05,
+      "loss": 1.3645,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 1.3599,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.185897435897436e-05,
+      "loss": 1.4028,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 1.3231,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1987179487179487e-05,
+      "loss": 1.3809,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 1.3292,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 1.3134,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 1.3903,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2243589743589746e-05,
+      "loss": 1.3395,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.3134,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2371794871794874e-05,
+      "loss": 1.5017,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 1.2739,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.25e-05,
+      "loss": 1.386,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 1.4265,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2628205128205129e-05,
+      "loss": 1.3041,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 1.3629,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2756410256410257e-05,
+      "loss": 1.3535,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 1.3041,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2884615384615386e-05,
+      "loss": 1.4734,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 1.4624,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3012820512820514e-05,
+      "loss": 1.3264,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.3737,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3141025641025642e-05,
+      "loss": 1.4304,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 1.3738,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 1.2598,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.3751,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3397435897435897e-05,
+      "loss": 1.3347,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 1.2162,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3525641025641028e-05,
+      "loss": 1.3395,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 1.298,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3653846153846156e-05,
+      "loss": 1.3773,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3717948717948718e-05,
+      "loss": 1.3239,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3782051282051283e-05,
+      "loss": 1.2872,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.3217,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3910256410256411e-05,
+      "loss": 1.3351,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 1.3041,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.403846153846154e-05,
+      "loss": 1.3232,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 1.4219,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 1.2932,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 1.36,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4294871794871796e-05,
+      "loss": 1.2953,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 1.4326,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 1.3699,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 1.3735,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4551282051282051e-05,
+      "loss": 1.3418,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 1.3584,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.467948717948718e-05,
+      "loss": 1.3235,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 1.3276,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.480769230769231e-05,
+      "loss": 1.353,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 1.3385,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4935897435897438e-05,
+      "loss": 1.3421,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.2881,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5064102564102565e-05,
+      "loss": 1.2928,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 1.3586,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5192307692307693e-05,
+      "loss": 1.3649,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5256410256410257e-05,
+      "loss": 1.3164,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5320512820512823e-05,
+      "loss": 1.4338,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.3561,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.544871794871795e-05,
+      "loss": 1.2131,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 1.2943,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.557692307692308e-05,
+      "loss": 1.3616,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 1.2881,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5705128205128205e-05,
+      "loss": 1.2296,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 1.2148,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 1.3488,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 1.208,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5961538461538465e-05,
+      "loss": 1.2733,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 1.3369,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6089743589743593e-05,
+      "loss": 1.2839,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.2318,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6217948717948718e-05,
+      "loss": 1.213,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 1.2682,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6346153846153847e-05,
+      "loss": 1.2832,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 1.3195,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6474358974358975e-05,
+      "loss": 1.3378,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 1.3651,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6602564102564103e-05,
+      "loss": 1.3002,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.2745,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.673076923076923e-05,
+      "loss": 1.4041,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 1.2755,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.685897435897436e-05,
+      "loss": 1.2831,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.2058,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.698717948717949e-05,
+      "loss": 1.3015,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 1.2931,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7115384615384617e-05,
+      "loss": 1.1782,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 1.2908,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7243589743589745e-05,
+      "loss": 1.2206,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 1.3981,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7371794871794873e-05,
+      "loss": 1.2977,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 1.1926,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.2159,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7564102564102566e-05,
+      "loss": 1.3795,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.762820512820513e-05,
+      "loss": 1.3236,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.1497,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.775641025641026e-05,
+      "loss": 1.3169,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 1.2899,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7884615384615387e-05,
+      "loss": 1.2401,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.2793,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8012820512820515e-05,
+      "loss": 1.3237,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 1.176,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8141025641025644e-05,
+      "loss": 1.3109,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 1.3493,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.826923076923077e-05,
+      "loss": 1.2767,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 1.2553,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8397435897435897e-05,
+      "loss": 1.2565,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.2408,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.852564102564103e-05,
+      "loss": 1.226,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8589743589743593e-05,
+      "loss": 1.2583,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8653846153846157e-05,
+      "loss": 1.3632,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 1.2781,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8782051282051282e-05,
+      "loss": 1.3274,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 1.195,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.891025641025641e-05,
+      "loss": 1.22,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.3887,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.903846153846154e-05,
+      "loss": 1.2174,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9102564102564106e-05,
+      "loss": 1.3389,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 1.2781,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.2186,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9294871794871796e-05,
+      "loss": 1.2379,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 1.1331,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9423076923076924e-05,
+      "loss": 1.252,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 1.3594,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9551282051282052e-05,
+      "loss": 1.2524,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 1.3156,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.967948717948718e-05,
+      "loss": 1.2629,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 1.2737,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980769230769231e-05,
+      "loss": 1.2256,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9871794871794873e-05,
+      "loss": 1.2905,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9935897435897437e-05,
+      "loss": 1.334,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 1.2061,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999999514514394e-05,
+      "loss": 1.2614,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999998058057616e-05,
+      "loss": 1.274,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999999563062981e-05,
+      "loss": 1.2833,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999992232231216e-05,
+      "loss": 1.3551,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999998786286216e-05,
+      "loss": 1.28,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999998252252306e-05,
+      "loss": 1.2844,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999976211214448e-05,
+      "loss": 1.2004,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999968928936924e-05,
+      "loss": 1.2607,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999960675691205e-05,
+      "loss": 1.23,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999995145147809e-05,
+      "loss": 1.248,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999994125629847e-05,
+      "loss": 1.2249,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999930090153335e-05,
+      "loss": 1.2305,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999917953043774e-05,
+      "loss": 1.2916,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999904844970963e-05,
+      "loss": 1.2137,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999890765936177e-05,
+      "loss": 1.2824,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999987571594078e-05,
+      "loss": 1.2623,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999859694986236e-05,
+      "loss": 1.2598,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99998427030741e-05,
+      "loss": 1.2288,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999824740206016e-05,
+      "loss": 1.2375,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999980580638374e-05,
+      "loss": 1.1125,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999785901609096e-05,
+      "loss": 1.1745,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999976502588403e-05,
+      "loss": 1.2539,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999974317921056e-05,
+      "loss": 1.2073,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999720361590812e-05,
+      "loss": 1.3292,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999696573027e-05,
+      "loss": 1.1682,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999671813521435e-05,
+      "loss": 1.2729,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999646083076517e-05,
+      "loss": 1.2719,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999961938169475e-05,
+      "loss": 1.29,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999591709378722e-05,
+      "loss": 1.2375,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999563066131124e-05,
+      "loss": 1.4084,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999533451954735e-05,
+      "loss": 1.2822,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999502866852427e-05,
+      "loss": 1.1852,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999471310827174e-05,
+      "loss": 1.237,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999943878388204e-05,
+      "loss": 1.1799,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999405286020184e-05,
+      "loss": 1.2569,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999370817244853e-05,
+      "loss": 1.3127,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999335377559403e-05,
+      "loss": 1.1614,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999298966967264e-05,
+      "loss": 1.3027,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999261585471982e-05,
+      "loss": 1.2117,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999223233077178e-05,
+      "loss": 1.2089,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999183909786583e-05,
+      "loss": 1.2022,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999914361560401e-05,
+      "loss": 1.2322,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999102350533376e-05,
+      "loss": 1.1844,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999060114578682e-05,
+      "loss": 1.1764,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999016907744034e-05,
+      "loss": 1.1861,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998972730033624e-05,
+      "loss": 1.184,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998927581451743e-05,
+      "loss": 1.2599,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998881462002778e-05,
+      "loss": 1.218,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99988343716912e-05,
+      "loss": 1.1906,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998786310521585e-05,
+      "loss": 1.2301,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998737278498602e-05,
+      "loss": 1.2634,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998687275627008e-05,
+      "loss": 1.2728,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998636301911657e-05,
+      "loss": 1.1976,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998584357357503e-05,
+      "loss": 1.1731,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999853144196959e-05,
+      "loss": 1.1258,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998477555753054e-05,
+      "loss": 1.2434,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998422698713126e-05,
+      "loss": 1.2517,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998366870855134e-05,
+      "loss": 1.2029,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998310072184495e-05,
+      "loss": 1.2244,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999825230270673e-05,
+      "loss": 1.1843,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998193562427443e-05,
+      "loss": 1.2578,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998133851352342e-05,
+      "loss": 1.2047,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999807316948722e-05,
+      "loss": 1.3118,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998011516837974e-05,
+      "loss": 1.2094,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997948893410585e-05,
+      "loss": 1.1706,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999788529921114e-05,
+      "loss": 1.24,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999782073424581e-05,
+      "loss": 1.2178,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999775519852086e-05,
+      "loss": 1.2419,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997688692042663e-05,
+      "loss": 1.1782,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999762121481767e-05,
+      "loss": 1.2758,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997552766852434e-05,
+      "loss": 1.3035,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99974833481536e-05,
+      "loss": 1.2098,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997412958727908e-05,
+      "loss": 1.2859,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997341598582197e-05,
+      "loss": 1.2694,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997269267723392e-05,
+      "loss": 1.2614,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997195966158518e-05,
+      "loss": 1.2582,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999712169389469e-05,
+      "loss": 1.207,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997046450939122e-05,
+      "loss": 1.2395,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999697023729912e-05,
+      "loss": 1.1708,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996893052982083e-05,
+      "loss": 1.1519,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99968148979955e-05,
+      "loss": 1.301,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996735772346973e-05,
+      "loss": 1.2487,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996655676044173e-05,
+      "loss": 1.2532,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996574609094887e-05,
+      "loss": 1.2397,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996492571506972e-05,
+      "loss": 1.1863,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996409563288404e-05,
+      "loss": 1.127,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996325584447243e-05,
+      "loss": 1.2609,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996240634991645e-05,
+      "loss": 1.3062,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999615471492985e-05,
+      "loss": 1.2803,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996067824270204e-05,
+      "loss": 1.1549,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999597996302115e-05,
+      "loss": 1.2313,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999589113119121e-05,
+      "loss": 1.207,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999580132878901e-05,
+      "loss": 1.329,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995710555823277e-05,
+      "loss": 1.2961,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999561881230282e-05,
+      "loss": 1.1976,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999552609823655e-05,
+      "loss": 1.1468,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995432413633462e-05,
+      "loss": 1.2053,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999533775850266e-05,
+      "loss": 1.1864,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995242132853333e-05,
+      "loss": 1.1464,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995145536694764e-05,
+      "loss": 1.2085,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995047970036334e-05,
+      "loss": 1.1394,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994949432887512e-05,
+      "loss": 1.1918,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994849925257874e-05,
+      "loss": 1.2629,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999474944715708e-05,
+      "loss": 1.1417,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999464799859488e-05,
+      "loss": 1.2192,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994545579581125e-05,
+      "loss": 1.2101,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994442190125765e-05,
+      "loss": 1.2289,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994337830238836e-05,
+      "loss": 1.3544,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999423249993047e-05,
+      "loss": 1.2465,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994126199210897e-05,
+      "loss": 1.1984,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999401892809044e-05,
+      "loss": 1.2253,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999391068657951e-05,
+      "loss": 1.2533,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993801474688615e-05,
+      "loss": 1.2062,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993691292428364e-05,
+      "loss": 1.2406,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999358013980946e-05,
+      "loss": 1.1375,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993468016842684e-05,
+      "loss": 1.1637,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993354923538926e-05,
+      "loss": 1.1803,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999324085990918e-05,
+      "loss": 1.2346,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993125825964502e-05,
+      "loss": 1.2682,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993009821716076e-05,
+      "loss": 1.2876,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992892847175155e-05,
+      "loss": 1.1754,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992774902353104e-05,
+      "loss": 1.1943,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992655987261375e-05,
+      "loss": 1.2911,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999253610191151e-05,
+      "loss": 1.2192,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992415246315155e-05,
+      "loss": 1.2455,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999229342048404e-05,
+      "loss": 1.2505,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992170624429998e-05,
+      "loss": 1.1658,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992046858164942e-05,
+      "loss": 1.1974,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9991922121700905e-05,
+      "loss": 1.2721,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999179641504999e-05,
+      "loss": 1.1895,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991669738224398e-05,
+      "loss": 1.2615,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991542091236438e-05,
+      "loss": 1.1655,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99914134740985e-05,
+      "loss": 1.2651,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991283886823075e-05,
+      "loss": 1.1722,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991153329422742e-05,
+      "loss": 1.2438,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991021801910177e-05,
+      "loss": 1.1722,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990889304298157e-05,
+      "loss": 1.1992,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999075583659954e-05,
+      "loss": 1.2811,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999062139882729e-05,
+      "loss": 1.2466,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999048599099446e-05,
+      "loss": 1.223,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990349613114195e-05,
+      "loss": 1.1334,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990212265199738e-05,
+      "loss": 1.2207,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990073947264426e-05,
+      "loss": 1.2119,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998993465932169e-05,
+      "loss": 1.1591,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989794401385054e-05,
+      "loss": 1.1831,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989653173468137e-05,
+      "loss": 1.1516,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998951097558465e-05,
+      "loss": 1.2415,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99893678077484e-05,
+      "loss": 1.2302,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998922366997329e-05,
+      "loss": 1.1455,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989078562273313e-05,
+      "loss": 1.193,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988932484662564e-05,
+      "loss": 1.2053,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988785437155222e-05,
+      "loss": 1.0864,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988637419765563e-05,
+      "loss": 1.1611,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988488432507963e-05,
+      "loss": 1.1761,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988338475396886e-05,
+      "loss": 1.0907,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988187548446895e-05,
+      "loss": 1.1719,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988035651672643e-05,
+      "loss": 1.1625,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998788278508888e-05,
+      "loss": 1.1853,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987728948710442e-05,
+      "loss": 1.1321,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987574142552274e-05,
+      "loss": 1.1558,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987418366629406e-05,
+      "loss": 1.1663,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987261620956964e-05,
+      "loss": 1.1891,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987103905550166e-05,
+      "loss": 1.226,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986945220424326e-05,
+      "loss": 1.1577,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986785565594848e-05,
+      "loss": 1.1859,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998662494107724e-05,
+      "loss": 1.2006,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998646334688709e-05,
+      "loss": 1.1802,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99863007830401e-05,
+      "loss": 1.2107,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998613724955205e-05,
+      "loss": 1.1548,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985972746438815e-05,
+      "loss": 1.1414,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985807273716366e-05,
+      "loss": 1.0688,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985640831400778e-05,
+      "loss": 1.1954,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985473419508205e-05,
+      "loss": 1.155,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998530503805491e-05,
+      "loss": 1.1527,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998513568705723e-05,
+      "loss": 1.2423,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984965366531624e-05,
+      "loss": 1.2116,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984794076494616e-05,
+      "loss": 1.2261,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984621816962843e-05,
+      "loss": 1.2208,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984448587953036e-05,
+      "loss": 1.1355,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984274389482005e-05,
+      "loss": 1.1721,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984099221566673e-05,
+      "loss": 1.2034,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983923084224047e-05,
+      "loss": 1.1565,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983745977471223e-05,
+      "loss": 1.2003,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983567901325404e-05,
+      "loss": 1.1377,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983388855803875e-05,
+      "loss": 1.2268,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983208840924028e-05,
+      "loss": 1.2673,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998302785670333e-05,
+      "loss": 1.1032,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998284590315937e-05,
+      "loss": 1.1699,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982662980309807e-05,
+      "loss": 1.1628,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982479088172403e-05,
+      "loss": 1.1627,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998229422676501e-05,
+      "loss": 1.2426,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982108396105584e-05,
+      "loss": 1.2571,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981921596212163e-05,
+      "loss": 1.2078,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981733827102884e-05,
+      "loss": 1.1895,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981545088795987e-05,
+      "loss": 1.129,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998135538130979e-05,
+      "loss": 1.1564,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998116470466272e-05,
+      "loss": 1.073,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998097305887328e-05,
+      "loss": 1.1334,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980780443960087e-05,
+      "loss": 1.1541,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980586859941846e-05,
+      "loss": 1.2915,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980392306837344e-05,
+      "loss": 1.26,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998019678466548e-05,
+      "loss": 1.1432,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980000293445232e-05,
+      "loss": 1.1304,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997980283319568e-05,
+      "loss": 1.1917,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979604403936005e-05,
+      "loss": 1.1315,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979405005685466e-05,
+      "loss": 1.05,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979204638463425e-05,
+      "loss": 1.1447,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979003302289336e-05,
+      "loss": 1.0516,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997880099718275e-05,
+      "loss": 1.2535,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997859772316331e-05,
+      "loss": 1.2588,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978393480250757e-05,
+      "loss": 1.1729,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978188268464912e-05,
+      "loss": 1.1652,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977982087825714e-05,
+      "loss": 1.0945,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997777493835317e-05,
+      "loss": 1.1545,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977566820067403e-05,
+      "loss": 1.1476,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977357732988616e-05,
+      "loss": 1.1178,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977147677137116e-05,
+      "loss": 1.1626,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976936652533288e-05,
+      "loss": 1.1594,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976724659197632e-05,
+      "loss": 1.2059,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997651169715073e-05,
+      "loss": 1.1937,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976297766413255e-05,
+      "loss": 1.2219,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976082867005985e-05,
+      "loss": 1.1376,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975866998949784e-05,
+      "loss": 1.1785,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997565016226561e-05,
+      "loss": 1.1384,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997543235697452e-05,
+      "loss": 1.2267,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997521358309766e-05,
+      "loss": 1.1583,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997499384065628e-05,
+      "loss": 1.142,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99747731296717e-05,
+      "loss": 1.1938,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974551450165366e-05,
+      "loss": 1.1313,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974328802158798e-05,
+      "loss": 1.1917,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974105185673613e-05,
+      "loss": 1.1346,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997388060073152e-05,
+      "loss": 1.18,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997365504735433e-05,
+      "loss": 1.2804,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973428525563948e-05,
+      "loss": 1.1578,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973201035382362e-05,
+      "loss": 1.1008,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972972576831656e-05,
+      "loss": 1.1259,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972743149934027e-05,
+      "loss": 1.1561,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972512754711738e-05,
+      "loss": 1.0809,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997228139118717e-05,
+      "loss": 1.1398,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997204905938278e-05,
+      "loss": 1.1864,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971815759321132e-05,
+      "loss": 1.1652,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971581491024873e-05,
+      "loss": 1.1552,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971346254516757e-05,
+      "loss": 1.0524,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997111004981962e-05,
+      "loss": 1.1689,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970872876956394e-05,
+      "loss": 1.2133,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970634735950117e-05,
+      "loss": 1.1274,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970395626823908e-05,
+      "loss": 1.1353,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970155549600978e-05,
+      "loss": 1.1259,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996991450430464e-05,
+      "loss": 1.1029,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9969672490958304e-05,
+      "loss": 1.1849,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9969429509585468e-05,
+      "loss": 1.1837,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996918556020972e-05,
+      "loss": 1.1681,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996894064285475e-05,
+      "loss": 1.2304,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996869475754434e-05,
+      "loss": 1.0649,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968447904302358e-05,
+      "loss": 1.0577,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968200083152784e-05,
+      "loss": 1.1058,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967951294119667e-05,
+      "loss": 1.1504,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967701537227175e-05,
+      "loss": 1.2434,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967450812499554e-05,
+      "loss": 1.1548,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996719911996115e-05,
+      "loss": 1.2022,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966946459636404e-05,
+      "loss": 1.264,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996669283154984e-05,
+      "loss": 1.1276,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966438235726094e-05,
+      "loss": 1.1183,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996618267218988e-05,
+      "loss": 1.1998,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965926140966017e-05,
+      "loss": 1.1462,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996566864207941e-05,
+      "loss": 1.2283,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965410175555064e-05,
+      "loss": 1.1744,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965150741418072e-05,
+      "loss": 1.1461,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996489033969363e-05,
+      "loss": 1.1331,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964628970407018e-05,
+      "loss": 1.0625,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964366633583614e-05,
+      "loss": 1.1767,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964103329248892e-05,
+      "loss": 1.1401,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963839057428417e-05,
+      "loss": 1.2111,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996357381814785e-05,
+      "loss": 1.2209,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963307611432944e-05,
+      "loss": 1.1337,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996304043730955e-05,
+      "loss": 1.2574,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962772295803602e-05,
+      "loss": 1.1162,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962503186941143e-05,
+      "loss": 1.2224,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99622331107483e-05,
+      "loss": 1.11,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961962067251298e-05,
+      "loss": 1.1621,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961690056476452e-05,
+      "loss": 1.2203,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961417078450177e-05,
+      "loss": 1.1406,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961143133198976e-05,
+      "loss": 1.1509,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996086822074945e-05,
+      "loss": 1.2285,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996059234112829e-05,
+      "loss": 1.1082,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960315494362286e-05,
+      "loss": 1.1619,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996003768047831e-05,
+      "loss": 1.1072,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959758899503355e-05,
+      "loss": 1.196,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995947915146447e-05,
+      "loss": 1.1397,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995919843638883e-05,
+      "loss": 1.1821,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995891675430369e-05,
+      "loss": 1.2067,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958634105236395e-05,
+      "loss": 1.1451,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99583504892144e-05,
+      "loss": 1.1829,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958065906265228e-05,
+      "loss": 1.1753,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957780356416525e-05,
+      "loss": 1.169,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957493839696013e-05,
+      "loss": 1.1772,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957206356131505e-05,
+      "loss": 1.2278,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956917905750926e-05,
+      "loss": 1.1738,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956628488582276e-05,
+      "loss": 1.1404,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995633810465366e-05,
+      "loss": 1.1478,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995604675399327e-05,
+      "loss": 1.0719,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99557544366294e-05,
+      "loss": 1.1943,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995546115259043e-05,
+      "loss": 1.1044,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955166901904838e-05,
+      "loss": 1.1198,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954871684601195e-05,
+      "loss": 1.1542,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954575500708164e-05,
+      "loss": 1.1077,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954278350254506e-05,
+      "loss": 1.1096,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995398023326907e-05,
+      "loss": 1.1707,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995368114978081e-05,
+      "loss": 1.1283,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953381099818756e-05,
+      "loss": 1.1108,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995308008341205e-05,
+      "loss": 1.1732,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952778100589912e-05,
+      "loss": 1.0897,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952475151381675e-05,
+      "loss": 1.1538,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952171235816747e-05,
+      "loss": 1.1431,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951866353924634e-05,
+      "loss": 1.1501,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951560505734948e-05,
+      "loss": 1.037,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951253691277376e-05,
+      "loss": 1.0983,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950945910581718e-05,
+      "loss": 1.1594,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950637163677854e-05,
+      "loss": 1.1017,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950327450595766e-05,
+      "loss": 1.149,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995001677136552e-05,
+      "loss": 1.1951,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949705126017286e-05,
+      "loss": 1.137,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949392514581322e-05,
+      "loss": 1.1669,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949078937087988e-05,
+      "loss": 1.1234,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948764393567722e-05,
+      "loss": 1.1311,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994844888405107e-05,
+      "loss": 1.1419,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994813240856867e-05,
+      "loss": 1.1769,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947814967151246e-05,
+      "loss": 1.1667,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947496559829622e-05,
+      "loss": 1.1453,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947177186634716e-05,
+      "loss": 1.1578,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946856847597535e-05,
+      "loss": 1.1513,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946535542749187e-05,
+      "loss": 1.1324,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946213272120865e-05,
+      "loss": 1.1576,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9945890035743866e-05,
+      "loss": 1.0467,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9945565833649574e-05,
+      "loss": 1.1425,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9945240665869465e-05,
+      "loss": 1.1838,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9944914532435113e-05,
+      "loss": 1.1371,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944587433378187e-05,
+      "loss": 1.1198,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944259368730443e-05,
+      "loss": 1.0657,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994393033852374e-05,
+      "loss": 1.1315,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943600342790022e-05,
+      "loss": 1.2111,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943269381561334e-05,
+      "loss": 1.0912,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994293745486981e-05,
+      "loss": 1.1855,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994260456274768e-05,
+      "loss": 1.1331,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9942270705227264e-05,
+      "loss": 1.1806,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941935882340976e-05,
+      "loss": 1.0673,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941600094121338e-05,
+      "loss": 1.1587,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994126334060094e-05,
+      "loss": 1.048,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994092562181249e-05,
+      "loss": 1.1904,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994058693778878e-05,
+      "loss": 1.0137,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940247288562683e-05,
+      "loss": 1.1085,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9939906674167192e-05,
+      "loss": 1.1707,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993956509463537e-05,
+      "loss": 1.1434,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993922255000039e-05,
+      "loss": 1.1833,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9938879040295508e-05,
+      "loss": 1.1975,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993853456555408e-05,
+      "loss": 1.1752,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993818912580955e-05,
+      "loss": 1.1182,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937842721095468e-05,
+      "loss": 1.2141,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993749535144546e-05,
+      "loss": 1.1884,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937147016893257e-05,
+      "loss": 1.1803,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993679771747268e-05,
+      "loss": 1.0891,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9936447453217646e-05,
+      "loss": 1.1545,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993609622416217e-05,
+      "loss": 1.0775,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935744030340347e-05,
+      "loss": 1.1726,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993539087178638e-05,
+      "loss": 1.014,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935036748534555e-05,
+      "loss": 1.157,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934681660619256e-05,
+      "loss": 1.1744,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993432560807497e-05,
+      "loss": 1.2236,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933968590936258e-05,
+      "loss": 1.17,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993361060923779e-05,
+      "loss": 1.1493,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993325166301433e-05,
+      "loss": 1.0795,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932891752300717e-05,
+      "loss": 1.1506,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932530877131912e-05,
+      "loss": 1.1274,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932169037542947e-05,
+      "loss": 1.1235,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931806233568957e-05,
+      "loss": 1.1245,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931442465245164e-05,
+      "loss": 1.171,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99310777326069e-05,
+      "loss": 1.1696,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930712035689576e-05,
+      "loss": 1.0602,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930345374528693e-05,
+      "loss": 1.1215,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992997774915986e-05,
+      "loss": 1.1329,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992960915961877e-05,
+      "loss": 1.1508,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992923960594121e-05,
+      "loss": 1.1246,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928869088163062e-05,
+      "loss": 1.1712,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928497606320308e-05,
+      "loss": 1.292,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928125160449018e-05,
+      "loss": 1.1736,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992775175058535e-05,
+      "loss": 1.0956,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992737737676556e-05,
+      "loss": 1.1415,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9927002039026002e-05,
+      "loss": 1.1169,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992662573740312e-05,
+      "loss": 1.1353,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9926248471933453e-05,
+      "loss": 1.2113,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992587024265363e-05,
+      "loss": 1.1418,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9925491049600382e-05,
+      "loss": 1.1292,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9925110892810515e-05,
+      "loss": 1.1155,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924729772320953e-05,
+      "loss": 1.15,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924347688168696e-05,
+      "loss": 1.0676,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923964640390846e-05,
+      "loss": 1.1549,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923580629024592e-05,
+      "loss": 1.114,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923195654107227e-05,
+      "loss": 1.132,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992280971567612e-05,
+      "loss": 1.1789,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992242281376876e-05,
+      "loss": 1.2488,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99220349484227e-05,
+      "loss": 1.1895,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921646119675606e-05,
+      "loss": 1.1805,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992125632756523e-05,
+      "loss": 1.1503,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9920865572129426e-05,
+      "loss": 1.212,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992047385340613e-05,
+      "loss": 1.1617,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9920081171433377e-05,
+      "loss": 1.1774,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9919687526249298e-05,
+      "loss": 1.2129,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991929291789211e-05,
+      "loss": 1.1968,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918897346400136e-05,
+      "loss": 1.1383,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918500811811778e-05,
+      "loss": 1.1426,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991810331416554e-05,
+      "loss": 1.1105,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991770485350002e-05,
+      "loss": 1.091,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99173054298539e-05,
+      "loss": 1.1868,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991690504326597e-05,
+      "loss": 1.0749,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9916503693775112e-05,
+      "loss": 1.177,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9916101381420285e-05,
+      "loss": 1.1486,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915698106240555e-05,
+      "loss": 1.0963,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915293868275083e-05,
+      "loss": 1.1685,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914888667563114e-05,
+      "loss": 1.0314,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914482504143996e-05,
+      "loss": 1.0635,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914075378057162e-05,
+      "loss": 1.1757,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9913667289342147e-05,
+      "loss": 1.1117,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9913258238038575e-05,
+      "loss": 1.1569,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991284822418616e-05,
+      "loss": 1.0492,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912437247824717e-05,
+      "loss": 1.1536,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912025308994146e-05,
+      "loss": 1.085,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9911612407734455e-05,
+      "loss": 1.0702,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9911198544085723e-05,
+      "loss": 1.1862,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910783718088142e-05,
+      "loss": 1.1195,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991036792978199e-05,
+      "loss": 1.1855,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9909951179207635e-05,
+      "loss": 1.1523,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990953346640555e-05,
+      "loss": 1.1907,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9909114791416283e-05,
+      "loss": 1.1786,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908695154280496e-05,
+      "loss": 1.2235,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990827455503893e-05,
+      "loss": 1.2109,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9907852993732425e-05,
+      "loss": 1.054,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9907430470401913e-05,
+      "loss": 1.1661,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990700698508842e-05,
+      "loss": 1.1604,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906582537833064e-05,
+      "loss": 1.0609,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990615712867706e-05,
+      "loss": 1.1174,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9905730757661712e-05,
+      "loss": 1.1265,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9905303424828418e-05,
+      "loss": 1.1294,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904875130218676e-05,
+      "loss": 1.1561,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904445873874068e-05,
+      "loss": 1.0902,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990401565583627e-05,
+      "loss": 1.0966,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903584476147066e-05,
+      "loss": 1.1642,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903152334848314e-05,
+      "loss": 1.1584,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902719231981975e-05,
+      "loss": 1.1362,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99022851675901e-05,
+      "loss": 1.1163,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901850141714843e-05,
+      "loss": 1.0901,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901414154398437e-05,
+      "loss": 1.1373,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9900977205683213e-05,
+      "loss": 1.1411,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9900539295611602e-05,
+      "loss": 1.2028,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9900100424226124e-05,
+      "loss": 1.1616,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9899660591569395e-05,
+      "loss": 1.1453,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9899219797684113e-05,
+      "loss": 1.1307,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9898778042613086e-05,
+      "loss": 1.0758,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98983353263992e-05,
+      "loss": 1.1724,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897891649085447e-05,
+      "loss": 1.118,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897447010714905e-05,
+      "loss": 1.111,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897001411330748e-05,
+      "loss": 1.129,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989655485097624e-05,
+      "loss": 1.056,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989610732969474e-05,
+      "loss": 1.0916,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989565884752971e-05,
+      "loss": 1.1547,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9895209404524684e-05,
+      "loss": 1.1055,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9894759000723308e-05,
+      "loss": 1.0406,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9894307636169312e-05,
+      "loss": 1.058,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9893855310906526e-05,
+      "loss": 1.1108,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989340202497887e-05,
+      "loss": 1.1716,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9892947778430352e-05,
+      "loss": 0.9804,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989249257130508e-05,
+      "loss": 1.1667,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9892036403647256e-05,
+      "loss": 1.1505,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989157927550117e-05,
+      "loss": 1.1414,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9891121186911207e-05,
+      "loss": 1.0709,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890662137921848e-05,
+      "loss": 1.0945,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890202128577664e-05,
+      "loss": 1.1938,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889741158923323e-05,
+      "loss": 1.0993,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988927922900358e-05,
+      "loss": 1.1702,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888816338863285e-05,
+      "loss": 1.1259,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888352488547397e-05,
+      "loss": 1.0201,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9887887678100938e-05,
+      "loss": 1.1182,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988742190756905e-05,
+      "loss": 1.181,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988695517699695e-05,
+      "loss": 1.1428,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9886487486429966e-05,
+      "loss": 1.1052,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98860188359135e-05,
+      "loss": 1.1783,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885549225493064e-05,
+      "loss": 1.1097,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885078655214252e-05,
+      "loss": 1.0359,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884607125122753e-05,
+      "loss": 1.1385,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988413463526436e-05,
+      "loss": 1.0746,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988366118568494e-05,
+      "loss": 1.1862,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988318677643047e-05,
+      "loss": 1.0946,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988271140754701e-05,
+      "loss": 1.1733,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9882235079080717e-05,
+      "loss": 1.1195,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9881757791077848e-05,
+      "loss": 1.1776,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9881279543584737e-05,
+      "loss": 1.0929,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9880800336647825e-05,
+      "loss": 1.1711,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988032017031364e-05,
+      "loss": 1.0829,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987983904462881e-05,
+      "loss": 1.1621,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9879356959640043e-05,
+      "loss": 1.0692,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878873915394154e-05,
+      "loss": 1.14,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878389911938042e-05,
+      "loss": 1.0953,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9877904949318704e-05,
+      "loss": 1.0759,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9877419027583227e-05,
+      "loss": 1.1564,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9876932146778796e-05,
+      "loss": 1.1257,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987644430695268e-05,
+      "loss": 1.0555,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9875955508152254e-05,
+      "loss": 1.1563,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987546575042497e-05,
+      "loss": 1.1478,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987497503381839e-05,
+      "loss": 1.0826,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987448335838016e-05,
+      "loss": 1.1187,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9873990724158014e-05,
+      "loss": 1.1232,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9873497131199793e-05,
+      "loss": 1.183,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987300257955342e-05,
+      "loss": 1.1347,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9872507069266914e-05,
+      "loss": 1.0392,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987201060038839e-05,
+      "loss": 1.171,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987151317296605e-05,
+      "loss": 1.1735,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871014787048197e-05,
+      "loss": 1.1515,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870515442683223e-05,
+      "loss": 1.1757,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870015139919606e-05,
+      "loss": 1.1301,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986951387880593e-05,
+      "loss": 1.1178,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9869011659390866e-05,
+      "loss": 1.0974,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9868508481723178e-05,
+      "loss": 1.0713,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9868004345851716e-05,
+      "loss": 1.1079,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9867499251825443e-05,
+      "loss": 1.0805,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9866993199693393e-05,
+      "loss": 1.0536,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9866486189504705e-05,
+      "loss": 1.0667,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.98659782213086e-05,
+      "loss": 1.1627,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9865469295154417e-05,
+      "loss": 1.1281,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986495941109156e-05,
+      "loss": 1.1214,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9864448569169536e-05,
+      "loss": 1.1503,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9863936769437956e-05,
+      "loss": 1.0291,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.98634240119465e-05,
+      "loss": 1.15,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986291029674497e-05,
+      "loss": 1.1556,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862395623883236e-05,
+      "loss": 1.091,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986187999341128e-05,
+      "loss": 1.0358,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9861363405379155e-05,
+      "loss": 1.1559,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9860845859837034e-05,
+      "loss": 1.1155,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9860327356835157e-05,
+      "loss": 1.0604,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985980789642388e-05,
+      "loss": 1.1299,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985928747865364e-05,
+      "loss": 1.0742,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985876610357496e-05,
+      "loss": 1.221,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985824377123847e-05,
+      "loss": 1.0817,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9857720481694887e-05,
+      "loss": 1.1013,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9857196234995016e-05,
+      "loss": 1.0942,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9856671031189765e-05,
+      "loss": 1.0883,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9856144870330133e-05,
+      "loss": 1.1739,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98556177524672e-05,
+      "loss": 1.0374,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985508967765215e-05,
+      "loss": 1.0438,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9854560645936262e-05,
+      "loss": 1.1382,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98540306573709e-05,
+      "loss": 1.1603,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9853499712007523e-05,
+      "loss": 1.1257,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9852967809897685e-05,
+      "loss": 1.1062,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9852434951093035e-05,
+      "loss": 1.1026,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9851901135645315e-05,
+      "loss": 1.1932,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985136636360635e-05,
+      "loss": 1.1304,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9850830635028063e-05,
+      "loss": 1.1202,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985029394996248e-05,
+      "loss": 1.116,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9849756308461705e-05,
+      "loss": 1.0961,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9849217710577945e-05,
+      "loss": 0.9974,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848678156363497e-05,
+      "loss": 1.0704,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848137645870745e-05,
+      "loss": 1.2251,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9847596179152176e-05,
+      "loss": 1.0676,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9847053756260363e-05,
+      "loss": 1.0745,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9846510377247974e-05,
+      "loss": 1.0875,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984596604216777e-05,
+      "loss": 1.1595,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845420751072604e-05,
+      "loss": 1.1027,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984487450401542e-05,
+      "loss": 1.1223,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984432730104926e-05,
+      "loss": 1.0547,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843779142227258e-05,
+      "loss": 1.1407,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984323002760263e-05,
+      "loss": 1.0831,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9842679957228706e-05,
+      "loss": 1.101,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9842128931158885e-05,
+      "loss": 1.0894,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9841576949446675e-05,
+      "loss": 1.1176,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984102401214567e-05,
+      "loss": 1.1161,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984047011930956e-05,
+      "loss": 1.0697,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9839915270992126e-05,
+      "loss": 1.0432,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9839359467247243e-05,
+      "loss": 1.1279,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9838802708128878e-05,
+      "loss": 1.1218,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983824499369109e-05,
+      "loss": 1.1144,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9837686323988032e-05,
+      "loss": 1.1398,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9837126699073948e-05,
+      "loss": 1.1222,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9836566119003175e-05,
+      "loss": 1.1108,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9836004583830146e-05,
+      "loss": 1.0989,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9835442093609383e-05,
+      "loss": 1.1324,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9834878648395507e-05,
+      "loss": 1.0933,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9834314248243218e-05,
+      "loss": 1.1382,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9833748893207326e-05,
+      "loss": 1.0773,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983318258334272e-05,
+      "loss": 1.1835,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9832615318704388e-05,
+      "loss": 1.1736,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983204709934741e-05,
+      "loss": 1.1196,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9831477925326962e-05,
+      "loss": 1.0837,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983090779669831e-05,
+      "loss": 1.0894,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98303367135168e-05,
+      "loss": 1.0737,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9829764675837894e-05,
+      "loss": 1.1132,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9829191683717133e-05,
+      "loss": 1.0663,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.982861773721015e-05,
+      "loss": 1.0469,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9828042836372677e-05,
+      "loss": 1.1469,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9827466981260534e-05,
+      "loss": 1.1265,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826890171929634e-05,
+      "loss": 1.16,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982631240843598e-05,
+      "loss": 1.1573,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982573369083568e-05,
+      "loss": 1.1327,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982515401918492e-05,
+      "loss": 1.1189,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9824573393539984e-05,
+      "loss": 1.2357,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9823991813957253e-05,
+      "loss": 1.1623,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982340928049319e-05,
+      "loss": 1.1534,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9822825793204364e-05,
+      "loss": 1.0902,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9822241352147426e-05,
+      "loss": 1.0808,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9821655957379128e-05,
+      "loss": 1.0757,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9821069608956307e-05,
+      "loss": 1.2148,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9820482306935892e-05,
+      "loss": 1.1049,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9819894051374917e-05,
+      "loss": 1.1022,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981930484233049e-05,
+      "loss": 1.1617,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981871467985983e-05,
+      "loss": 1.0608,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9818123564020234e-05,
+      "loss": 1.1478,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9817531494869105e-05,
+      "loss": 1.1381,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9816938472463923e-05,
+      "loss": 1.1049,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9816344496862272e-05,
+      "loss": 1.0679,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9815749568121828e-05,
+      "loss": 1.1333,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9815153686300352e-05,
+      "loss": 1.0894,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9814556851455707e-05,
+      "loss": 1.0668,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981395906364584e-05,
+      "loss": 1.1056,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.98133603229288e-05,
+      "loss": 1.0517,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9812760629362714e-05,
+      "loss": 1.1814,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9812159983005823e-05,
+      "loss": 1.1479,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981155838391643e-05,
+      "loss": 1.1421,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9810955832152968e-05,
+      "loss": 1.1215,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9810352327773935e-05,
+      "loss": 1.1381,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9809747870837927e-05,
+      "loss": 1.1282,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9809142461403635e-05,
+      "loss": 1.164,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9808536099529844e-05,
+      "loss": 1.0949,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9807928785275433e-05,
+      "loss": 1.1212,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980732051869937e-05,
+      "loss": 1.1589,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980671129986071e-05,
+      "loss": 1.0545,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9806101128818613e-05,
+      "loss": 1.1183,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9805490005632323e-05,
+      "loss": 1.0898,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9804877930361173e-05,
+      "loss": 1.145,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.98042649030646e-05,
+      "loss": 1.1104,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9803650923802126e-05,
+      "loss": 1.0837,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9803035992633366e-05,
+      "loss": 1.0634,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980242010961803e-05,
+      "loss": 1.0133,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9801803274815915e-05,
+      "loss": 1.1351,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9801185488286916e-05,
+      "loss": 1.1149,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9800566750091018e-05,
+      "loss": 1.0975,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9799947060288297e-05,
+      "loss": 1.0488,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9799326418938924e-05,
+      "loss": 1.0776,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9798704826103165e-05,
+      "loss": 1.1203,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979808228184137e-05,
+      "loss": 1.0989,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979745878621399e-05,
+      "loss": 1.1806,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9796834339281557e-05,
+      "loss": 1.0767,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979620894110471e-05,
+      "loss": 1.0455,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979558259174418e-05,
+      "loss": 1.1222,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979495529126077e-05,
+      "loss": 1.0678,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9794327039715395e-05,
+      "loss": 1.1393,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979369783716906e-05,
+      "loss": 1.134,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979306768368285e-05,
+      "loss": 1.0991,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979243657931796e-05,
+      "loss": 1.0538,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9791804524135663e-05,
+      "loss": 1.1349,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979117151819733e-05,
+      "loss": 1.0748,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979053756156443e-05,
+      "loss": 1.0701,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978990265429851e-05,
+      "loss": 1.1379,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9789266796461222e-05,
+      "loss": 1.1156,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9788629988114306e-05,
+      "loss": 1.0952,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9787992229319594e-05,
+      "loss": 1.1768,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9787353520139007e-05,
+      "loss": 0.9994,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978671386063457e-05,
+      "loss": 1.111,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9786073250868384e-05,
+      "loss": 1.1315,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978543169090265e-05,
+      "loss": 1.051,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9784789180799674e-05,
+      "loss": 1.1277,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9784145720621827e-05,
+      "loss": 1.0367,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9783501310431596e-05,
+      "loss": 1.064,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9782855950291542e-05,
+      "loss": 1.052,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9782209640264344e-05,
+      "loss": 1.1367,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978156238041274e-05,
+      "loss": 1.0713,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9780914170799587e-05,
+      "loss": 1.2008,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9780265011487822e-05,
+      "loss": 1.1462,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977961490254047e-05,
+      "loss": 1.0842,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9778963844020668e-05,
+      "loss": 1.0445,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977831183599162e-05,
+      "loss": 1.0357,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977765887851664e-05,
+      "loss": 1.086,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9777004971659126e-05,
+      "loss": 1.0951,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977635011548257e-05,
+      "loss": 1.1713,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977569431005056e-05,
+      "loss": 1.0811,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9775037555426772e-05,
+      "loss": 1.0576,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9774379851674973e-05,
+      "loss": 1.1962,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9773721198859024e-05,
+      "loss": 1.0135,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977306159704288e-05,
+      "loss": 1.1129,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9772401046290584e-05,
+      "loss": 1.1235,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9771739546666278e-05,
+      "loss": 1.071,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9771077098234187e-05,
+      "loss": 1.133,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9770413701058632e-05,
+      "loss": 1.0609,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9769749355204034e-05,
+      "loss": 1.1251,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9769084060734895e-05,
+      "loss": 1.1724,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976841781771581e-05,
+      "loss": 1.1067,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9767750626211475e-05,
+      "loss": 1.1043,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9767082486286667e-05,
+      "loss": 1.0365,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9766413398006267e-05,
+      "loss": 1.0958,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9765743361435234e-05,
+      "loss": 1.117,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9765072376638633e-05,
+      "loss": 1.0972,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9764400443681607e-05,
+      "loss": 1.0995,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976372756262941e-05,
+      "loss": 1.0845,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9763053733547367e-05,
+      "loss": 1.0658,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9762378956500906e-05,
+      "loss": 1.1666,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.976170323155555e-05,
+      "loss": 1.1362,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.976102655877691e-05,
+      "loss": 1.2136,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.976034893823069e-05,
+      "loss": 1.1158,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759670369982675e-05,
+      "loss": 1.1451,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975899085409876e-05,
+      "loss": 1.1373,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9758310390644932e-05,
+      "loss": 1.0884,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9757628979687247e-05,
+      "loss": 1.0555,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9756946621291873e-05,
+      "loss": 1.0605,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975626331552507e-05,
+      "loss": 1.0302,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975557906245318e-05,
+      "loss": 1.1545,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9754893862142643e-05,
+      "loss": 1.0613,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975420771465999e-05,
+      "loss": 1.1065,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9753520620071846e-05,
+      "loss": 1.1515,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975283257844492e-05,
+      "loss": 1.1794,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9752143589846027e-05,
+      "loss": 1.0418,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975145365434206e-05,
+      "loss": 1.1431,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9750762772000014e-05,
+      "loss": 1.0663,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975007094288697e-05,
+      "loss": 1.0973,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9749378167070097e-05,
+      "loss": 1.1007,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9748684444616673e-05,
+      "loss": 1.0208,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9747989775594044e-05,
+      "loss": 1.1844,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974729416006967e-05,
+      "loss": 1.0508,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974659759811109e-05,
+      "loss": 1.0696,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745900089785936e-05,
+      "loss": 1.0625,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745201635161938e-05,
+      "loss": 1.1598,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9744502234306907e-05,
+      "loss": 1.1504,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9743801887288762e-05,
+      "loss": 1.0845,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9743100594175497e-05,
+      "loss": 1.0613,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9742398355035212e-05,
+      "loss": 1.144,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9741695169936087e-05,
+      "loss": 1.0856,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9740991038946404e-05,
+      "loss": 1.0529,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9740285962134527e-05,
+      "loss": 1.1111,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973957993956892e-05,
+      "loss": 1.1314,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9738872971318134e-05,
+      "loss": 1.076,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9738165057450817e-05,
+      "loss": 1.1066,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97374561980357e-05,
+      "loss": 1.1777,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9736746393141617e-05,
+      "loss": 1.0063,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9736035642837486e-05,
+      "loss": 1.0241,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9735323947192317e-05,
+      "loss": 1.1227,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9734611306275214e-05,
+      "loss": 1.0916,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9733897720155377e-05,
+      "loss": 1.1495,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9733183188902086e-05,
+      "loss": 1.1497,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9732467712584723e-05,
+      "loss": 1.1354,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973175129127276e-05,
+      "loss": 1.0768,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973103392503576e-05,
+      "loss": 1.1191,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9730315613943372e-05,
+      "loss": 1.1508,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9729596358065347e-05,
+      "loss": 1.0974,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.972887615747152e-05,
+      "loss": 1.1384,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9728155012231825e-05,
+      "loss": 1.0881,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.972743292241628e-05,
+      "loss": 1.0298,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9726709888094994e-05,
+      "loss": 1.0842,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9725985909338176e-05,
+      "loss": 1.1602,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972526098621612e-05,
+      "loss": 1.1841,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972453511879922e-05,
+      "loss": 1.0262,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972380830715795e-05,
+      "loss": 1.1339,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972308055136288e-05,
+      "loss": 1.061,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9722351851484677e-05,
+      "loss": 1.2159,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9721622207594092e-05,
+      "loss": 1.0927,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9720891619761974e-05,
+      "loss": 1.1182,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972016008805926e-05,
+      "loss": 1.0714,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9719427612556982e-05,
+      "loss": 1.041,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9718694193326256e-05,
+      "loss": 1.1716,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9717959830438302e-05,
+      "loss": 1.1179,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9717224523964417e-05,
+      "loss": 1.0815,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9716488273976006e-05,
+      "loss": 1.0565,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9715751080544544e-05,
+      "loss": 1.0904,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971501294374162e-05,
+      "loss": 1.0761,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9714273863638908e-05,
+      "loss": 1.0381,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971353384030816e-05,
+      "loss": 1.0564,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9712792873821235e-05,
+      "loss": 1.0562,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9712050964250083e-05,
+      "loss": 1.1262,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9711308111666736e-05,
+      "loss": 1.0677,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9710564316143323e-05,
+      "loss": 1.0956,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9709819577752064e-05,
+      "loss": 1.0476,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9709073896565276e-05,
+      "loss": 1.0786,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9708327272655362e-05,
+      "loss": 1.0648,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9707579706094807e-05,
+      "loss": 1.1148,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9706831196956208e-05,
+      "loss": 1.0285,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970608174531224e-05,
+      "loss": 1.0891,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9705331351235673e-05,
+      "loss": 1.0869,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970458001479937e-05,
+      "loss": 0.9998,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9703827736076276e-05,
+      "loss": 1.074,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9703074515139445e-05,
+      "loss": 1.0494,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9702320352062005e-05,
+      "loss": 1.1219,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9701565246917184e-05,
+      "loss": 1.0534,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970080919977831e-05,
+      "loss": 1.0983,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9700052210718775e-05,
+      "loss": 1.045,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.96992942798121e-05,
+      "loss": 1.1019,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969853540713186e-05,
+      "loss": 1.1376,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9697775592751757e-05,
+      "loss": 1.1062,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9697014836745552e-05,
+      "loss": 1.0403,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969625313918712e-05,
+      "loss": 1.1156,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9695490500150418e-05,
+      "loss": 1.0706,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9694726919709496e-05,
+      "loss": 1.1118,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9693962397938495e-05,
+      "loss": 1.0586,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9693196934911652e-05,
+      "loss": 1.0325,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9692430530703282e-05,
+      "loss": 1.0796,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969166318538781e-05,
+      "loss": 1.1833,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9690894899039735e-05,
+      "loss": 1.1219,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969012567173366e-05,
+      "loss": 1.0236,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9689355503544277e-05,
+      "loss": 1.0835,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9688584394546364e-05,
+      "loss": 1.1867,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.968781234481479e-05,
+      "loss": 1.1923,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9687039354424528e-05,
+      "loss": 1.0515,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9686265423450624e-05,
+      "loss": 1.1061,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9685490551968233e-05,
+      "loss": 1.1544,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9684714740052584e-05,
+      "loss": 0.9973,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968393798777901e-05,
+      "loss": 1.1263,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9683160295222934e-05,
+      "loss": 1.1499,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968238166245986e-05,
+      "loss": 1.1268,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9681602089565403e-05,
+      "loss": 1.0662,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9680821576615247e-05,
+      "loss": 1.0925,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968004012368518e-05,
+      "loss": 1.1085,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967925773085108e-05,
+      "loss": 1.0457,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967847439818892e-05,
+      "loss": 1.0022,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9677690125774752e-05,
+      "loss": 1.0665,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9676904913684725e-05,
+      "loss": 1.0945,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967611876199509e-05,
+      "loss": 1.1211,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967533167078217e-05,
+      "loss": 1.078,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96745436401224e-05,
+      "loss": 1.1416,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9673754670092283e-05,
+      "loss": 1.1538,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967296476076844e-05,
+      "loss": 1.1021,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9672173912227556e-05,
+      "loss": 1.0651,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9671382124546427e-05,
+      "loss": 1.0275,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967058939780193e-05,
+      "loss": 1.1598,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966979573207104e-05,
+      "loss": 1.1184,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966900112743082e-05,
+      "loss": 1.1533,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9668205583958418e-05,
+      "loss": 1.0493,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966740910173108e-05,
+      "loss": 1.1071,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9666611680826153e-05,
+      "loss": 1.0822,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9665813321321054e-05,
+      "loss": 1.0769,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9665014023293298e-05,
+      "loss": 1.1811,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9664213786820502e-05,
+      "loss": 1.076,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9663412611980368e-05,
+      "loss": 1.1116,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9662610498850684e-05,
+      "loss": 1.0485,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966180744750933e-05,
+      "loss": 1.1415,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9661003458034288e-05,
+      "loss": 1.1235,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660198530503613e-05,
+      "loss": 1.0673,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965939266499547e-05,
+      "loss": 1.0977,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96585858615881e-05,
+      "loss": 1.0869,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9657778120359848e-05,
+      "loss": 1.1033,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9656969441389137e-05,
+      "loss": 1.0536,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965615982475449e-05,
+      "loss": 1.0675,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965534927053452e-05,
+      "loss": 1.0729,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9654537778807924e-05,
+      "loss": 1.0585,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9653725349653503e-05,
+      "loss": 1.151,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9652911983150135e-05,
+      "loss": 1.0845,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9652097679376802e-05,
+      "loss": 1.1407,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965128243841256e-05,
+      "loss": 1.1087,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965046626033658e-05,
+      "loss": 1.0818,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96496491452281e-05,
+      "loss": 1.0631,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9648831093166464e-05,
+      "loss": 1.119,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9648012104231106e-05,
+      "loss": 1.1462,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964719217850154e-05,
+      "loss": 1.119,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964637131605738e-05,
+      "loss": 1.0862,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9645549516978337e-05,
+      "loss": 1.0572,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9644726781344197e-05,
+      "loss": 1.1038,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643903109234844e-05,
+      "loss": 1.1324,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964307850073026e-05,
+      "loss": 1.0964,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9642252955910513e-05,
+      "loss": 1.1259,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964142647485576e-05,
+      "loss": 1.0963,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964059905764624e-05,
+      "loss": 1.0396,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9639770704362305e-05,
+      "loss": 1.0902,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9638941415084384e-05,
+      "loss": 1.0881,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9638111189892994e-05,
+      "loss": 1.0714,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963728002886875e-05,
+      "loss": 1.059,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9636447932092354e-05,
+      "loss": 1.1356,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9635614899644604e-05,
+      "loss": 0.9661,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963478093160638e-05,
+      "loss": 1.1246,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9633946028058662e-05,
+      "loss": 1.1846,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9633110189082515e-05,
+      "loss": 1.1707,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9632273414759093e-05,
+      "loss": 1.0631,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963143570516965e-05,
+      "loss": 1.0325,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9630597060395526e-05,
+      "loss": 1.1132,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9629757480518144e-05,
+      "loss": 1.0502,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962891696561903e-05,
+      "loss": 1.1279,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9628075515779796e-05,
+      "loss": 1.1542,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962723313108214e-05,
+      "loss": 1.095,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962638981160786e-05,
+      "loss": 1.0391,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9625545557438837e-05,
+      "loss": 1.011,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9624700368657045e-05,
+      "loss": 1.122,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9623854245344555e-05,
+      "loss": 1.1432,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9623007187583518e-05,
+      "loss": 1.146,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962215919545618e-05,
+      "loss": 1.1457,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962131026904488e-05,
+      "loss": 1.0659,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962046040843205e-05,
+      "loss": 1.0977,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.96196096137002e-05,
+      "loss": 1.089,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961875788493195e-05,
+      "loss": 1.0481,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9617905222209998e-05,
+      "loss": 1.1251,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961705162561713e-05,
+      "loss": 1.0674,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961619709523623e-05,
+      "loss": 1.0938,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961534163115027e-05,
+      "loss": 1.0787,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9614485233442316e-05,
+      "loss": 0.9917,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9613627902195523e-05,
+      "loss": 1.1278,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961276963749313e-05,
+      "loss": 1.1401,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9611910439418472e-05,
+      "loss": 1.0283,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9611050308054982e-05,
+      "loss": 1.1066,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9610189243486167e-05,
+      "loss": 1.1062,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9609327245795642e-05,
+      "loss": 1.1189,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.96084643150671e-05,
+      "loss": 1.0797,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9607600451384327e-05,
+      "loss": 1.0418,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960673565483121e-05,
+      "loss": 1.1254,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960586992549171e-05,
+      "loss": 1.1218,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9605003263449887e-05,
+      "loss": 1.1632,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9604135668789897e-05,
+      "loss": 1.0668,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960326714159598e-05,
+      "loss": 1.1294,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9602397681952462e-05,
+      "loss": 1.1265,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9601527289943768e-05,
+      "loss": 1.087,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9600655965654413e-05,
+      "loss": 1.0826,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9599783709169e-05,
+      "loss": 1.0921,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959891052057222e-05,
+      "loss": 1.0471,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9598036399948855e-05,
+      "loss": 1.0052,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9597161347383783e-05,
+      "loss": 1.0975,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959628536296197e-05,
+      "loss": 1.1465,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959540844676847e-05,
+      "loss": 1.183,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959453059888843e-05,
+      "loss": 1.0793,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9593651819407084e-05,
+      "loss": 1.0586,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959277210840976e-05,
+      "loss": 1.1373,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959189146598188e-05,
+      "loss": 1.0995,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9591009892208945e-05,
+      "loss": 1.1364,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9590127387176556e-05,
+      "loss": 1.116,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95892439509704e-05,
+      "loss": 1.1155,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9588359583676263e-05,
+      "loss": 1.1101,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9587474285380007e-05,
+      "loss": 1.1053,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9586588056167595e-05,
+      "loss": 1.0432,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9585700896125074e-05,
+      "loss": 1.0041,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958481280533859e-05,
+      "loss": 1.0406,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9583923783894373e-05,
+      "loss": 1.0731,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958303383187874e-05,
+      "loss": 1.0743,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9582142949378104e-05,
+      "loss": 1.0552,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9581251136478974e-05,
+      "loss": 1.0625,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9580358393267933e-05,
+      "loss": 1.0754,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9579464719831668e-05,
+      "loss": 1.0182,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9578570116256953e-05,
+      "loss": 1.0444,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9577674582630653e-05,
+      "loss": 1.092,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9576778119039715e-05,
+      "loss": 1.0957,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957588072557119e-05,
+      "loss": 1.1066,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957498240231221e-05,
+      "loss": 1.0682,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957408314935e-05,
+      "loss": 1.0956,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9573182966771874e-05,
+      "loss": 1.0013,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9572281854665233e-05,
+      "loss": 1.0541,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957137981311758e-05,
+      "loss": 1.1351,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95704768422165e-05,
+      "loss": 1.0364,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9569572942049664e-05,
+      "loss": 1.0701,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956866811270484e-05,
+      "loss": 1.0617,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9567762354269882e-05,
+      "loss": 1.109,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9566855666832743e-05,
+      "loss": 1.0419,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9565948050481456e-05,
+      "loss": 0.9855,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9565039505304145e-05,
+      "loss": 1.0718,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9564130031389032e-05,
+      "loss": 1.0971,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956321962882442e-05,
+      "loss": 1.1128,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956230829769871e-05,
+      "loss": 1.1036,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956139603810039e-05,
+      "loss": 1.0743,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9560482850118035e-05,
+      "loss": 1.1293,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9559568733840317e-05,
+      "loss": 1.1068,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9558653689355983e-05,
+      "loss": 1.1514,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9557737716753898e-05,
+      "loss": 1.0962,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9556820816122987e-05,
+      "loss": 1.0459,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9555902987552283e-05,
+      "loss": 1.0988,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9554984231130908e-05,
+      "loss": 1.0892,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9554064546948064e-05,
+      "loss": 1.0516,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9553143935093052e-05,
+      "loss": 1.2318,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9552222395655262e-05,
+      "loss": 1.0218,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9551299928724177e-05,
+      "loss": 1.0823,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9550376534389355e-05,
+      "loss": 0.936,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9549452212740468e-05,
+      "loss": 1.0922,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9548526963867253e-05,
+      "loss": 1.1351,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9547600787859557e-05,
+      "loss": 1.0493,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9546673684807303e-05,
+      "loss": 1.1031,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9545745654800515e-05,
+      "loss": 1.1453,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95448166979293e-05,
+      "loss": 1.1799,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9543886814283858e-05,
+      "loss": 1.0934,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9542956003954477e-05,
+      "loss": 1.0526,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9542024267031537e-05,
+      "loss": 1.0569,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9541091603605508e-05,
+      "loss": 1.1053,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9540158013766947e-05,
+      "loss": 1.0992,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95392234976065e-05,
+      "loss": 1.1426,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953828805521491e-05,
+      "loss": 1.1047,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9537351686683003e-05,
+      "loss": 1.0482,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9536414392101702e-05,
+      "loss": 1.0922,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9535476171562014e-05,
+      "loss": 1.1268,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9534537025155034e-05,
+      "loss": 1.0193,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9533596952971955e-05,
+      "loss": 1.1172,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9532655955104052e-05,
+      "loss": 1.0746,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9531714031642698e-05,
+      "loss": 1.0777,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9530771182679342e-05,
+      "loss": 1.0752,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9529827408305542e-05,
+      "loss": 1.0917,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952888270861293e-05,
+      "loss": 1.0969,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527937083693233e-05,
+      "loss": 1.1231,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9526990533638272e-05,
+      "loss": 1.1007,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952604305853995e-05,
+      "loss": 1.1328,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952509465849027e-05,
+      "loss": 1.064,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9524145333581315e-05,
+      "loss": 1.0461,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952319508390526e-05,
+      "loss": 1.0386,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9522243909554375e-05,
+      "loss": 1.0258,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9521291810621018e-05,
+      "loss": 1.0259,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952033878719763e-05,
+      "loss": 1.1556,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951938483937675e-05,
+      "loss": 1.0018,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518429967251e-05,
+      "loss": 1.0692,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95174741709131e-05,
+      "loss": 1.0177,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9516517450455853e-05,
+      "loss": 1.0259,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9515559805972156e-05,
+      "loss": 1.2578,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951460123755499e-05,
+      "loss": 1.0517,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9513641745297427e-05,
+      "loss": 1.1361,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9512681329292635e-05,
+      "loss": 1.1818,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951171998963387e-05,
+      "loss": 1.0709,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951075772641447e-05,
+      "loss": 1.0077,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9509794539727874e-05,
+      "loss": 1.1439,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95088304296676e-05,
+      "loss": 1.0322,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9507865396327262e-05,
+      "loss": 1.0743,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950689943980056e-05,
+      "loss": 1.0562,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9505932560181286e-05,
+      "loss": 1.0911,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9504964757563322e-05,
+      "loss": 1.0691,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950399603204064e-05,
+      "loss": 0.9924,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.95030263837073e-05,
+      "loss": 1.0041,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950205581265745e-05,
+      "loss": 1.0854,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9501084318985335e-05,
+      "loss": 1.0516,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9500111902785276e-05,
+      "loss": 1.0397,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94991385641517e-05,
+      "loss": 1.075,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949816430317911e-05,
+      "loss": 1.0287,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9497189119962105e-05,
+      "loss": 1.0662,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9496213014595374e-05,
+      "loss": 1.0533,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9495235987173693e-05,
+      "loss": 1.0989,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949425803779193e-05,
+      "loss": 1.0854,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949327916654504e-05,
+      "loss": 1.101,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9492299373528066e-05,
+      "loss": 1.099,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949131865883614e-05,
+      "loss": 1.0571,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94903370225645e-05,
+      "loss": 0.9268,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948935446480845e-05,
+      "loss": 1.1169,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94883709856634e-05,
+      "loss": 1.0805,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948738658522483e-05,
+      "loss": 1.0918,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9486401263588335e-05,
+      "loss": 1.1521,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485415020849583e-05,
+      "loss": 1.1014,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9484427857104335e-05,
+      "loss": 1.0225,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9483439772448444e-05,
+      "loss": 1.0508,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9482450766977846e-05,
+      "loss": 1.1201,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9481460840788573e-05,
+      "loss": 1.0584,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9480469993976744e-05,
+      "loss": 1.0706,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9479478226638565e-05,
+      "loss": 1.1163,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947848553887034e-05,
+      "loss": 0.8561,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947749193076845e-05,
+      "loss": 1.063,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9476497402429372e-05,
+      "loss": 1.1,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9475501953949674e-05,
+      "loss": 1.1113,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947450558542601e-05,
+      "loss": 1.0403,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9473508296955126e-05,
+      "loss": 1.0883,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9472510088633857e-05,
+      "loss": 1.08,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9471510960559122e-05,
+      "loss": 1.0962,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9470510912827932e-05,
+      "loss": 1.0359,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9469509945537395e-05,
+      "loss": 1.0724,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9468508058784703e-05,
+      "loss": 1.0655,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9467505252667126e-05,
+      "loss": 1.1216,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9466501527282045e-05,
+      "loss": 1.1605,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9465496882726913e-05,
+      "loss": 1.069,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9464491319099278e-05,
+      "loss": 1.0077,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946348483649678e-05,
+      "loss": 1.1516,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946247743501714e-05,
+      "loss": 1.0846,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9461469114758184e-05,
+      "loss": 1.095,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946045987581781e-05,
+      "loss": 1.1098,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9459449718294008e-05,
+      "loss": 1.0599,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9458438642284867e-05,
+      "loss": 1.0764,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945742664788856e-05,
+      "loss": 1.0619,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945641373520335e-05,
+      "loss": 1.1179,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9455399904327585e-05,
+      "loss": 1.1697,
+      "step": 1376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9454385155359704e-05,
+      "loss": 1.109,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945336948839824e-05,
+      "loss": 1.0916,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9452352903541807e-05,
+      "loss": 1.1206,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9451335400889114e-05,
+      "loss": 1.1249,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9450316980538963e-05,
+      "loss": 1.0899,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944929764259023e-05,
+      "loss": 0.9851,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9448277387141896e-05,
+      "loss": 1.0143,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9447256214293026e-05,
+      "loss": 1.0319,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9446234124142768e-05,
+      "loss": 1.0784,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9445211116790365e-05,
+      "loss": 1.0657,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9444187192335152e-05,
+      "loss": 1.1078,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9443162350876544e-05,
+      "loss": 1.029,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9442136592514054e-05,
+      "loss": 1.1205,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944110991734728e-05,
+      "loss": 1.0588,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944008232547591e-05,
+      "loss": 1.0877,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9439053816999717e-05,
+      "loss": 1.0355,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9438024392018566e-05,
+      "loss": 1.1493,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9436994050632415e-05,
+      "loss": 1.1191,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9435962792941306e-05,
+      "loss": 1.1258,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9434930619045367e-05,
+      "loss": 1.0953,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9433897529044824e-05,
+      "loss": 1.0233,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9432863523039986e-05,
+      "loss": 1.073,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9431828601131252e-05,
+      "loss": 1.0523,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9430792763419105e-05,
+      "loss": 1.1205,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942975601000413e-05,
+      "loss": 1.0479,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942871834098699e-05,
+      "loss": 1.0787,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9427679756468435e-05,
+      "loss": 1.0308,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9426640256549313e-05,
+      "loss": 1.0747,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942559984133056e-05,
+      "loss": 1.0884,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9424558510913186e-05,
+      "loss": 1.1462,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942351626539831e-05,
+      "loss": 1.062,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9422473104887133e-05,
+      "loss": 1.1477,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9421429029480937e-05,
+      "loss": 1.1525,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9420384039281103e-05,
+      "loss": 1.04,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941933813438909e-05,
+      "loss": 1.0464,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941829131490646e-05,
+      "loss": 1.0367,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9417243580934852e-05,
+      "loss": 1.0609,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9416194932576e-05,
+      "loss": 1.0114,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941514536993172e-05,
+      "loss": 1.097,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941409489310393e-05,
+      "loss": 1.0342,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9413043502194616e-05,
+      "loss": 1.0333,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9411991197305878e-05,
+      "loss": 1.0999,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9410937978539886e-05,
+      "loss": 1.0677,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9409883845998905e-05,
+      "loss": 1.0762,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9408828799785286e-05,
+      "loss": 1.1024,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9407772840001473e-05,
+      "loss": 1.1239,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9406715966749998e-05,
+      "loss": 1.0788,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9405658180133477e-05,
+      "loss": 1.0245,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940459948025462e-05,
+      "loss": 1.1089,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9403539867216226e-05,
+      "loss": 1.0062,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9402479341121173e-05,
+      "loss": 1.0507,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9401417902072447e-05,
+      "loss": 0.9147,
+      "step": 1428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94003555501731e-05,
+      "loss": 1.1154,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9399292285526286e-05,
+      "loss": 1.1063,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939822810823525e-05,
+      "loss": 1.1602,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939716301840331e-05,
+      "loss": 1.0652,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939609701613389e-05,
+      "loss": 1.0963,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9395030101530504e-05,
+      "loss": 1.0327,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939396227469673e-05,
+      "loss": 1.1395,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939289353573626e-05,
+      "loss": 1.0881,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9391823884752865e-05,
+      "loss": 1.061,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9390753321850404e-05,
+      "loss": 1.0445,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9389681847132826e-05,
+      "loss": 1.0804,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938860946070417e-05,
+      "loss": 1.0854,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9387536162668556e-05,
+      "loss": 1.1241,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93864619531302e-05,
+      "loss": 1.0529,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9385386832193412e-05,
+      "loss": 1.0666,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9384310799962575e-05,
+      "loss": 1.037,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9383233856542167e-05,
+      "loss": 1.0939,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9382156002036764e-05,
+      "loss": 1.0589,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9381077236551017e-05,
+      "loss": 1.0816,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9379997560189677e-05,
+      "loss": 1.1288,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378916973057567e-05,
+      "loss": 1.0734,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937783547525962e-05,
+      "loss": 1.13,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9376753066900843e-05,
+      "loss": 1.0588,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9375669748086326e-05,
+      "loss": 1.0956,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937458551892127e-05,
+      "loss": 1.0868,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937350037951094e-05,
+      "loss": 0.9885,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9372414329960705e-05,
+      "loss": 1.0226,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9371327370376018e-05,
+      "loss": 0.9889,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9370239500862413e-05,
+      "loss": 1.1577,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936915072152553e-05,
+      "loss": 1.0101,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9368061032471074e-05,
+      "loss": 0.9827,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936697043380486e-05,
+      "loss": 1.0637,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9365878925632777e-05,
+      "loss": 1.0362,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936478650806081e-05,
+      "loss": 1.0619,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9363693181195028e-05,
+      "loss": 1.1014,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936259894514159e-05,
+      "loss": 0.9707,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9361503800006748e-05,
+      "loss": 1.0327,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9360407745896828e-05,
+      "loss": 1.0909,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935931078291826e-05,
+      "loss": 1.1082,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9358212911177556e-05,
+      "loss": 1.0434,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9357114130781314e-05,
+      "loss": 1.02,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935601444183622e-05,
+      "loss": 1.1605,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9354913844449062e-05,
+      "loss": 1.0999,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935381233872669e-05,
+      "loss": 1.1213,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9352709924776064e-05,
+      "loss": 1.0774,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935160660270423e-05,
+      "loss": 1.0551,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935050237261831e-05,
+      "loss": 1.1257,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.934939723462552e-05,
+      "loss": 1.1072,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9348291188833176e-05,
+      "loss": 1.1026,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9347184235348663e-05,
+      "loss": 1.0668,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9346076374279467e-05,
+      "loss": 1.087,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9344967605733154e-05,
+      "loss": 1.0679,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.934385792981739e-05,
+      "loss": 0.9923,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934274734663991e-05,
+      "loss": 1.104,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9341635856308558e-05,
+      "loss": 1.1084,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934052345893125e-05,
+      "loss": 1.0621,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9339410154616006e-05,
+      "loss": 1.1535,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9338295943470915e-05,
+      "loss": 1.1169,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9337180825604166e-05,
+      "loss": 1.0743,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9336064801124034e-05,
+      "loss": 1.0379,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9334947870138887e-05,
+      "loss": 1.1658,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933383003275717e-05,
+      "loss": 0.9703,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933271128908742e-05,
+      "loss": 1.0612,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933159163923827e-05,
+      "loss": 1.103,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933047108331843e-05,
+      "loss": 1.1209,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9329349621436708e-05,
+      "loss": 1.13,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932822725370199e-05,
+      "loss": 1.068,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9327103980223255e-05,
+      "loss": 1.1168,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932597980110957e-05,
+      "loss": 1.0557,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932485471647009e-05,
+      "loss": 1.1427,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932372872641406e-05,
+      "loss": 1.1417,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9322601831050804e-05,
+      "loss": 1.0997,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9321474030489747e-05,
+      "loss": 1.0856,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9320345324840396e-05,
+      "loss": 1.0131,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9319215714212337e-05,
+      "loss": 1.0632,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9318085198715257e-05,
+      "loss": 1.0311,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9316953778458925e-05,
+      "loss": 1.1323,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93158214535532e-05,
+      "loss": 1.1218,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9314688224108025e-05,
+      "loss": 1.1715,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9313554090233436e-05,
+      "loss": 1.0403,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9312419052039556e-05,
+      "loss": 1.1128,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9311283109636586e-05,
+      "loss": 1.024,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931014626313483e-05,
+      "loss": 1.051,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9309008512644668e-05,
+      "loss": 1.0885,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9307869858276575e-05,
+      "loss": 1.0356,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.930673030014111e-05,
+      "loss": 1.1423,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9305589838348925e-05,
+      "loss": 1.0486,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.930444847301075e-05,
+      "loss": 1.0029,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9303306204237406e-05,
+      "loss": 1.0453,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9302163032139813e-05,
+      "loss": 1.0251,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9301018956828966e-05,
+      "loss": 1.0591,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9299873978415947e-05,
+      "loss": 1.0959,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9298728097011932e-05,
+      "loss": 1.0146,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9297581312728187e-05,
+      "loss": 1.1133,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929643362567606e-05,
+      "loss": 1.0467,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929528503596698e-05,
+      "loss": 1.1144,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9294135543712483e-05,
+      "loss": 1.0632,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929298514902418e-05,
+      "loss": 1.1027,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9291833852013763e-05,
+      "loss": 1.0873,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929068165279303e-05,
+      "loss": 1.0885,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9289528551473843e-05,
+      "loss": 1.0848,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.928837454816818e-05,
+      "loss": 1.0188,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.928721964298808e-05,
+      "loss": 1.0199,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9286063836045687e-05,
+      "loss": 1.0934,
+      "step": 1532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9284907127453222e-05,
+      "loss": 1.0775,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9283749517323e-05,
+      "loss": 1.1431,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9282591005767423e-05,
+      "loss": 1.0902,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928143159289898e-05,
+      "loss": 1.0165,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9280271278830246e-05,
+      "loss": 1.0471,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927911006367388e-05,
+      "loss": 1.0242,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9277947947542635e-05,
+      "loss": 1.1048,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927678493054935e-05,
+      "loss": 1.1227,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9275621012806955e-05,
+      "loss": 1.0785,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9274456194428454e-05,
+      "loss": 1.0733,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9273290475526956e-05,
+      "loss": 0.9593,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9272123856215643e-05,
+      "loss": 1.0892,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927095633660779e-05,
+      "loss": 1.0553,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9269787916816764e-05,
+      "loss": 0.9749,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9268618596956014e-05,
+      "loss": 1.0898,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9267448377139074e-05,
+      "loss": 1.0212,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9266277257479577e-05,
+      "loss": 1.1342,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9265105238091227e-05,
+      "loss": 1.0903,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926393231908783e-05,
+      "loss": 1.0316,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9262758500583265e-05,
+      "loss": 1.069,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9261583782691514e-05,
+      "loss": 1.1851,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9260408165526638e-05,
+      "loss": 1.149,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925923164920278e-05,
+      "loss": 1.1191,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9258054233834184e-05,
+      "loss": 1.078,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925687591953517e-05,
+      "loss": 1.0609,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9255696706420147e-05,
+      "loss": 1.097,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9254516594603618e-05,
+      "loss": 1.0952,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9253335584200165e-05,
+      "loss": 1.0294,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925215367532446e-05,
+      "loss": 1.0813,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9250970868091268e-05,
+      "loss": 1.1456,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924978716261543e-05,
+      "loss": 1.0371,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924860255901188e-05,
+      "loss": 1.017,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924741705739565e-05,
+      "loss": 1.0344,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9246230657881834e-05,
+      "loss": 1.1162,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924504336058564e-05,
+      "loss": 1.1107,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9243855165622345e-05,
+      "loss": 0.9647,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9242666073107324e-05,
+      "loss": 1.0591,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9241476083156026e-05,
+      "loss": 1.0006,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9240285195884004e-05,
+      "loss": 1.1816,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9239093411406885e-05,
+      "loss": 1.0872,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9237900729840394e-05,
+      "loss": 1.0934,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9236707151300326e-05,
+      "loss": 1.0461,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9235512675902583e-05,
+      "loss": 0.9953,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9234317303763145e-05,
+      "loss": 1.0452,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923312103499807e-05,
+      "loss": 1.0558,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9231923869723528e-05,
+      "loss": 1.0166,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9230725808055747e-05,
+      "loss": 1.0899,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922952685011106e-05,
+      "loss": 1.0911,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922832699600588e-05,
+      "loss": 1.0746,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9227126245856716e-05,
+      "loss": 1.0486,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9225924599780148e-05,
+      "loss": 1.0232,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922472205789286e-05,
+      "loss": 1.153,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922351862031161e-05,
+      "loss": 1.0754,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9222314287153255e-05,
+      "loss": 1.0589,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9221109058534726e-05,
+      "loss": 1.0801,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9219902934573048e-05,
+      "loss": 1.072,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9218695915385337e-05,
+      "loss": 1.1033,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9217488001088784e-05,
+      "loss": 1.0764,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921627919180068e-05,
+      "loss": 1.1256,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9215069487638396e-05,
+      "loss": 1.1098,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9213858888719387e-05,
+      "loss": 1.0909,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.92126473951612e-05,
+      "loss": 0.9531,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921143500708147e-05,
+      "loss": 1.1063,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921022172459791e-05,
+      "loss": 0.9979,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920900754782834e-05,
+      "loss": 1.0884,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920779247689064e-05,
+      "loss": 1.0743,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9206576511902794e-05,
+      "loss": 0.9921,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9205359652982868e-05,
+      "loss": 1.1117,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920414190024902e-05,
+      "loss": 1.0814,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9202923253819482e-05,
+      "loss": 1.0089,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920170371381259e-05,
+      "loss": 1.0701,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920048328034675e-05,
+      "loss": 1.0283,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9199261953540467e-05,
+      "loss": 1.1077,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9198039733512326e-05,
+      "loss": 1.0417,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9196816620381004e-05,
+      "loss": 1.0893,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9195592614265262e-05,
+      "loss": 1.1147,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9194367715283945e-05,
+      "loss": 1.0851,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9193141923555984e-05,
+      "loss": 1.001,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9191915239200406e-05,
+      "loss": 1.1351,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919068766233632e-05,
+      "loss": 1.1019,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918945919308291e-05,
+      "loss": 1.043,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9188229831559468e-05,
+      "loss": 1.0849,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9186999577885353e-05,
+      "loss": 1.0682,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9185768432180026e-05,
+      "loss": 1.0444,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918453639456302e-05,
+      "loss": 1.0524,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9183303465153972e-05,
+      "loss": 1.0515,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918206964407259e-05,
+      "loss": 1.0471,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9180834931438673e-05,
+      "loss": 1.1033,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917959932737211e-05,
+      "loss": 0.9856,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917836283199288e-05,
+      "loss": 1.067,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9177125445421032e-05,
+      "loss": 1.0344,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917588716777672e-05,
+      "loss": 1.018,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9174647999180176e-05,
+      "loss": 1.1182,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917340793975172e-05,
+      "loss": 1.0522,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9172166989611756e-05,
+      "loss": 1.0925,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917092514888078e-05,
+      "loss": 1.1188,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916968241767937e-05,
+      "loss": 1.1359,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9168438796128193e-05,
+      "loss": 1.0735,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9167194284347993e-05,
+      "loss": 1.0662,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9165948882459623e-05,
+      "loss": 1.0401,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9164702590583994e-05,
+      "loss": 1.1216,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9163455408842123e-05,
+      "loss": 1.0663,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916220733735511e-05,
+      "loss": 0.9798,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9160958376244138e-05,
+      "loss": 1.14,
+      "step": 1636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9159708525630474e-05,
+      "loss": 1.1133,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9158457785635478e-05,
+      "loss": 1.0528,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915720615638059e-05,
+      "loss": 1.0546,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915595363798735e-05,
+      "loss": 1.068,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915470023057736e-05,
+      "loss": 1.0947,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915344593427233e-05,
+      "loss": 1.0595,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9152190749194047e-05,
+      "loss": 1.0959,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9150934675464384e-05,
+      "loss": 1.0189,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9149677713205307e-05,
+      "loss": 1.0557,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9148419862538858e-05,
+      "loss": 1.077,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9147161123587174e-05,
+      "loss": 1.0796,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9145901496472474e-05,
+      "loss": 1.0556,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9144640981317064e-05,
+      "loss": 1.1206,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9143379578243335e-05,
+      "loss": 1.11,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914211728737377e-05,
+      "loss": 1.0898,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914085410883093e-05,
+      "loss": 1.0513,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9139590042737466e-05,
+      "loss": 1.064,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9138325089216118e-05,
+      "loss": 1.1079,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9137059248389706e-05,
+      "loss": 1.0824,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913579252038114e-05,
+      "loss": 1.0339,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9134524905313422e-05,
+      "loss": 1.0872,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9133256403309627e-05,
+      "loss": 1.0837,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913198701449292e-05,
+      "loss": 1.0909,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913071673898656e-05,
+      "loss": 1.0432,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9129445576913886e-05,
+      "loss": 1.1044,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912817352839833e-05,
+      "loss": 0.9953,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9126900593563394e-05,
+      "loss": 1.0479,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9125626772532683e-05,
+      "loss": 1.0085,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912435206542988e-05,
+      "loss": 1.0626,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9123076472378753e-05,
+      "loss": 0.9478,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912179999350316e-05,
+      "loss": 1.0045,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9120522628927047e-05,
+      "loss": 1.0621,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9119244378774437e-05,
+      "loss": 1.0688,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9117965243169444e-05,
+      "loss": 1.1151,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9116685222236277e-05,
+      "loss": 0.9456,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9115404316099212e-05,
+      "loss": 1.103,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9114122524882628e-05,
+      "loss": 1.0364,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9112839848710978e-05,
+      "loss": 1.0808,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9111556287708813e-05,
+      "loss": 0.9171,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9110271842000755e-05,
+      "loss": 1.0942,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9108986511711527e-05,
+      "loss": 1.0786,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9107700296965926e-05,
+      "loss": 1.0765,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910641319788884e-05,
+      "loss": 1.0551,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910512521460525e-05,
+      "loss": 1.1129,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9103836347240205e-05,
+      "loss": 1.1083,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9102546595918857e-05,
+      "loss": 1.0694,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9101255960766436e-05,
+      "loss": 0.9764,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9099964441908258e-05,
+      "loss": 1.1064,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9098672039469723e-05,
+      "loss": 1.0574,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9097378753576327e-05,
+      "loss": 1.0485,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9096084584353635e-05,
+      "loss": 1.1125,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9094789531927315e-05,
+      "loss": 1.0938,
+      "step": 1688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.909349359642311e-05,
+      "loss": 1.1224,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.909219677796685e-05,
+      "loss": 1.0215,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9090899076684452e-05,
+      "loss": 1.063,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9089600492701926e-05,
+      "loss": 1.1254,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9088301026145353e-05,
+      "loss": 1.0251,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908700067714091e-05,
+      "loss": 1.1453,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9085699445814856e-05,
+      "loss": 0.931,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9084397332293537e-05,
+      "loss": 1.0281,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908309433670339e-05,
+      "loss": 1.0271,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9081790459170926e-05,
+      "loss": 1.0497,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9080485699822748e-05,
+      "loss": 1.039,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9079180058785547e-05,
+      "loss": 1.0834,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9077873536186094e-05,
+      "loss": 0.9946,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9076566132151255e-05,
+      "loss": 1.0884,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9075257846807964e-05,
+      "loss": 1.11,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907394868028326e-05,
+      "loss": 1.0604,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907263863270426e-05,
+      "loss": 1.0028,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9071327704198163e-05,
+      "loss": 1.1284,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9070015894892253e-05,
+      "loss": 1.0369,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906870320491391e-05,
+      "loss": 1.0234,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906738963439059e-05,
+      "loss": 1.0439,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9066075183449835e-05,
+      "loss": 1.1139,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906475985221927e-05,
+      "loss": 1.0787,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9063443640826624e-05,
+      "loss": 1.1255,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9062126549399687e-05,
+      "loss": 1.1208,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906080857806634e-05,
+      "loss": 0.9931,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9059489726954568e-05,
+      "loss": 1.0582,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905816999619242e-05,
+      "loss": 1.0705,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9056849385908032e-05,
+      "loss": 1.1039,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9055527896229642e-05,
+      "loss": 1.0478,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905420552728556e-05,
+      "loss": 1.0658,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905288227920418e-05,
+      "loss": 1.0305,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9051558152113986e-05,
+      "loss": 1.0834,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9050233146143554e-05,
+      "loss": 1.0017,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9048907261421533e-05,
+      "loss": 1.0672,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9047580498076663e-05,
+      "loss": 1.0968,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904625285623777e-05,
+      "loss": 1.0108,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904492433603376e-05,
+      "loss": 1.0746,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9043594937593633e-05,
+      "loss": 1.071,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904226466104647e-05,
+      "loss": 1.0979,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9040933506521436e-05,
+      "loss": 1.0375,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903960147414778e-05,
+      "loss": 1.0573,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903826856405484e-05,
+      "loss": 1.0592,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903693477637204e-05,
+      "loss": 1.0349,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9035600111228885e-05,
+      "loss": 1.1132,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9034264568754967e-05,
+      "loss": 1.0225,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9032928149079962e-05,
+      "loss": 1.0514,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9031590852333637e-05,
+      "loss": 1.0625,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903025267864583e-05,
+      "loss": 1.0768,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9028913628146487e-05,
+      "loss": 1.0126,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.902757370096562e-05,
+      "loss": 0.996,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.902623289723333e-05,
+      "loss": 1.021,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9024891217079803e-05,
+      "loss": 1.1089,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.902354866063532e-05,
+      "loss": 1.0536,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9022205228030234e-05,
+      "loss": 1.0444,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9020860919394992e-05,
+      "loss": 1.0258,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9019515734860117e-05,
+      "loss": 1.063,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9018169674556228e-05,
+      "loss": 1.0108,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901682273861402e-05,
+      "loss": 1.08,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901547492716428e-05,
+      "loss": 1.0814,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9014126240337876e-05,
+      "loss": 1.0327,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9012776678265756e-05,
+      "loss": 1.093,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9011426241078967e-05,
+      "loss": 1.1198,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9010074928908624e-05,
+      "loss": 1.059,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9008722741885943e-05,
+      "loss": 1.015,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900736968014221e-05,
+      "loss": 1.0811,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900601574380881e-05,
+      "loss": 1.0922,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9004660933017208e-05,
+      "loss": 1.0262,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9003305247898946e-05,
+      "loss": 1.0587,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900194868858566e-05,
+      "loss": 1.0149,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9000591255209066e-05,
+      "loss": 1.0571,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8999232947900968e-05,
+      "loss": 1.0988,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8997873766793255e-05,
+      "loss": 1.0859,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.89965137120179e-05,
+      "loss": 0.863,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8995152783706958e-05,
+      "loss": 1.0747,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899379098199257e-05,
+      "loss": 1.0897,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8992428307006968e-05,
+      "loss": 1.1023,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899106475888246e-05,
+      "loss": 1.0984,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8989700337751446e-05,
+      "loss": 1.0392,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.89883350437464e-05,
+      "loss": 1.012,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8986968876999898e-05,
+      "loss": 1.0474,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8985601837644586e-05,
+      "loss": 0.978,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.89842339258132e-05,
+      "loss": 1.0428,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8982865141638557e-05,
+      "loss": 1.1219,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8981495485253565e-05,
+      "loss": 1.0181,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8980124956791216e-05,
+      "loss": 1.0986,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897875355638458e-05,
+      "loss": 0.992,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8977381284166818e-05,
+      "loss": 1.1471,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8976008140271175e-05,
+      "loss": 1.1202,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897463412483098e-05,
+      "loss": 1.067,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897325923797964e-05,
+      "loss": 0.9851,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897188347985066e-05,
+      "loss": 1.0933,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8970506850577615e-05,
+      "loss": 1.1031,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896912935029418e-05,
+      "loss": 1.0825,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.89677509791341e-05,
+      "loss": 1.0488,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896637173723121e-05,
+      "loss": 1.0808,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8964991624719434e-05,
+      "loss": 1.0918,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8963610641732777e-05,
+      "loss": 1.1014,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8962228788405327e-05,
+      "loss": 1.0823,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8960846064871257e-05,
+      "loss": 1.0714,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.895946247126483e-05,
+      "loss": 1.0252,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8958078007720387e-05,
+      "loss": 1.001,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.895669267437235e-05,
+      "loss": 1.0418,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.895530647135524e-05,
+      "loss": 1.0862,
+      "step": 1792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8953919398803644e-05,
+      "loss": 1.0994,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8952531456852248e-05,
+      "loss": 1.0411,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8951142645635818e-05,
+      "loss": 0.999,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8949752965289197e-05,
+      "loss": 1.0355,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894836241594733e-05,
+      "loss": 1.0872,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894697099774523e-05,
+      "loss": 1.0042,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8945578710817995e-05,
+      "loss": 1.1075,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894418555530082e-05,
+      "loss": 1.0688,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8942791531328968e-05,
+      "loss": 1.0522,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.89413966390378e-05,
+      "loss": 1.0981,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8940000878562758e-05,
+      "loss": 1.1041,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8938604250039362e-05,
+      "loss": 1.0615,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8937206753603223e-05,
+      "loss": 1.0774,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8935808389390032e-05,
+      "loss": 1.0261,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8934409157535566e-05,
+      "loss": 1.0419,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.893300905817569e-05,
+      "loss": 1.0231,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8931608091446347e-05,
+      "loss": 0.9565,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8930206257483566e-05,
+      "loss": 1.0293,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8928803556423464e-05,
+      "loss": 1.0699,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8927399988402233e-05,
+      "loss": 1.0503,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892599555355616e-05,
+      "loss": 1.0463,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8924590252021614e-05,
+      "loss": 1.0665,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8923184083935044e-05,
+      "loss": 0.9824,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8921777049432985e-05,
+      "loss": 1.0861,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892036914865205e-05,
+      "loss": 1.0472,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8918960381728947e-05,
+      "loss": 1.1243,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8917550748800466e-05,
+      "loss": 0.9854,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8916140250003475e-05,
+      "loss": 1.071,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8914728885474928e-05,
+      "loss": 1.0768,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891331665535187e-05,
+      "loss": 1.0715,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891190355977142e-05,
+      "loss": 1.0475,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8910489598870784e-05,
+      "loss": 0.9723,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8909074772787254e-05,
+      "loss": 1.0402,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8907659081658214e-05,
+      "loss": 1.1359,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8906242525621112e-05,
+      "loss": 1.0202,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8904825104813497e-05,
+      "loss": 0.9954,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8903406819372997e-05,
+      "loss": 1.0053,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8901987669437322e-05,
+      "loss": 1.1599,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8900567655144267e-05,
+      "loss": 1.0246,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889914677663171e-05,
+      "loss": 0.9876,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889772503403762e-05,
+      "loss": 1.0975,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8896302427500042e-05,
+      "loss": 1.0719,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.88948789571571e-05,
+      "loss": 1.1036,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8893454623147017e-05,
+      "loss": 1.0607,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889202942560809e-05,
+      "loss": 1.0454,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.88906033646787e-05,
+      "loss": 0.9691,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8889176440497312e-05,
+      "loss": 1.1494,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8887748653202478e-05,
+      "loss": 1.0494,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8886320002932835e-05,
+      "loss": 1.0466,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8884890489827097e-05,
+      "loss": 1.1353,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8883460114024066e-05,
+      "loss": 1.0345,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8882028875662627e-05,
+      "loss": 1.0126,
+      "step": 1844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.888059677488175e-05,
+      "loss": 1.0341,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8879163811820493e-05,
+      "loss": 1.0559,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8877729986617983e-05,
+      "loss": 1.0778,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8876295299413445e-05,
+      "loss": 1.0635,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8874859750346185e-05,
+      "loss": 1.0856,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8873423339555584e-05,
+      "loss": 1.0919,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8871986067181124e-05,
+      "loss": 1.0515,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8870547933362352e-05,
+      "loss": 0.9354,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8869108938238904e-05,
+      "loss": 0.9706,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.886766908195051e-05,
+      "loss": 1.0356,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8866228364636976e-05,
+      "loss": 1.0922,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8864786786438187e-05,
+      "loss": 1.0801,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8863344347494114e-05,
+      "loss": 1.063,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.886190104794482e-05,
+      "loss": 1.0507,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8860456887930442e-05,
+      "loss": 1.0513,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8859011867591203e-05,
+      "loss": 1.1073,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885756598706741e-05,
+      "loss": 1.0357,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885611924649946e-05,
+      "loss": 1.0574,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8854671646027816e-05,
+      "loss": 1.0382,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885322318579305e-05,
+      "loss": 0.9965,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8851773865935786e-05,
+      "loss": 1.0671,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8850323686596766e-05,
+      "loss": 1.1245,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884887264791679e-05,
+      "loss": 1.0301,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8847420750036748e-05,
+      "loss": 1.0369,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8845967993097618e-05,
+      "loss": 1.0109,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884451437724046e-05,
+      "loss": 1.0435,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8843059902606413e-05,
+      "loss": 1.0671,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8841604569336702e-05,
+      "loss": 1.0276,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8840148377572638e-05,
+      "loss": 1.0854,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883869132745561e-05,
+      "loss": 1.0232,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.88372334191271e-05,
+      "loss": 0.9362,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883577465272866e-05,
+      "loss": 1.0146,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8834315028401934e-05,
+      "loss": 1.0244,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8832854546288642e-05,
+      "loss": 1.1171,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8831393206530605e-05,
+      "loss": 1.1146,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8829931009269707e-05,
+      "loss": 1.0609,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882846795464792e-05,
+      "loss": 1.07,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882700404280731e-05,
+      "loss": 1.0377,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8825539273890016e-05,
+      "loss": 1.0195,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8824073648038258e-05,
+      "loss": 0.9794,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882260716539435e-05,
+      "loss": 0.9943,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882113982610068e-05,
+      "loss": 1.0263,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8819671630299726e-05,
+      "loss": 1.0463,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881820257813404e-05,
+      "loss": 1.0746,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8816732669746268e-05,
+      "loss": 0.9778,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8815261905279133e-05,
+      "loss": 1.0593,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881379028487544e-05,
+      "loss": 1.0261,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8812317808678075e-05,
+      "loss": 0.9733,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8810844476830025e-05,
+      "loss": 1.0788,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8809370289474327e-05,
+      "loss": 1.0462,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.880789524675414e-05,
+      "loss": 0.9929,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8806419348812673e-05,
+      "loss": 1.0202,
+      "step": 1896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8804942595793236e-05,
+      "loss": 1.086,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8803464987839217e-05,
+      "loss": 1.0265,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8801986525094087e-05,
+      "loss": 0.992,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.88005072077014e-05,
+      "loss": 1.0944,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.87990270358048e-05,
+      "loss": 1.0677,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8797546009547996e-05,
+      "loss": 1.0881,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8796064129074797e-05,
+      "loss": 1.0996,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879458139452909e-05,
+      "loss": 1.0615,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879309780605485e-05,
+      "loss": 1.1033,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8791613363796118e-05,
+      "loss": 1.0721,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879012806789703e-05,
+      "loss": 1.0985,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8788641918501817e-05,
+      "loss": 1.0648,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8787154915754766e-05,
+      "loss": 0.9885,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8785667059800264e-05,
+      "loss": 1.035,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878417835078278e-05,
+      "loss": 1.1074,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8782688788846865e-05,
+      "loss": 1.0359,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878119837413715e-05,
+      "loss": 0.9865,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877970710679834e-05,
+      "loss": 1.0552,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8778214986975253e-05,
+      "loss": 1.0834,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877672201481275e-05,
+      "loss": 1.0816,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8775228190455805e-05,
+      "loss": 1.072,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877373351404946e-05,
+      "loss": 1.037,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8772237985738845e-05,
+      "loss": 1.0465,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8770741605669173e-05,
+      "loss": 1.0699,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8769244373985734e-05,
+      "loss": 1.0192,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876774629083391e-05,
+      "loss": 1.1028,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8766247356359157e-05,
+      "loss": 1.0711,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8764747570707017e-05,
+      "loss": 1.0876,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8763246934023122e-05,
+      "loss": 1.0959,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8761745446453167e-05,
+      "loss": 1.0458,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8760243108142954e-05,
+      "loss": 1.0366,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875873991923835e-05,
+      "loss": 1.0869,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8757235879885306e-05,
+      "loss": 1.1157,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875573099022987e-05,
+      "loss": 1.1351,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8754225250418158e-05,
+      "loss": 0.9731,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8752718660596367e-05,
+      "loss": 1.0525,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875121122091079e-05,
+      "loss": 0.997,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8749702931507797e-05,
+      "loss": 1.0691,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874819379253383e-05,
+      "loss": 1.1022,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874668380413543e-05,
+      "loss": 1.082,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8745172966459204e-05,
+      "loss": 1.0025,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8743661279651856e-05,
+      "loss": 1.0746,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874214874386017e-05,
+      "loss": 1.0723,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8740635359231e-05,
+      "loss": 0.9661,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8739121125911297e-05,
+      "loss": 1.0381,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8737606044048086e-05,
+      "loss": 1.0732,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873609011378848e-05,
+      "loss": 0.9917,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873457333527967e-05,
+      "loss": 1.0627,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8733055708668928e-05,
+      "loss": 0.9989,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873153723410362e-05,
+      "loss": 1.0349,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8730017911731172e-05,
+      "loss": 1.0543,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8728497741699115e-05,
+      "loss": 1.1025,
+      "step": 1948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8726976724155056e-05,
+      "loss": 1.1454,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872545485924667e-05,
+      "loss": 1.0585,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872393214712174e-05,
+      "loss": 1.0499,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8722408587928104e-05,
+      "loss": 1.1131,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.87208841818137e-05,
+      "loss": 1.0079,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8719358928926546e-05,
+      "loss": 1.0802,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.871783282941474e-05,
+      "loss": 1.0159,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8716305883426456e-05,
+      "loss": 1.0161,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8714778091109963e-05,
+      "loss": 1.0477,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.87132494526136e-05,
+      "loss": 1.0578,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8711719968085795e-05,
+      "loss": 1.0876,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8710189637675055e-05,
+      "loss": 1.0588,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8708658461529975e-05,
+      "loss": 1.0279,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8707126439799225e-05,
+      "loss": 1.0576,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8705593572631562e-05,
+      "loss": 1.0836,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870405986017582e-05,
+      "loss": 1.0953,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8702525302580915e-05,
+      "loss": 1.046,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8700989899995857e-05,
+      "loss": 0.9061,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8699453652569725e-05,
+      "loss": 1.0895,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8697916560451682e-05,
+      "loss": 1.0521,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.869637862379098e-05,
+      "loss": 1.1339,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.869483984273694e-05,
+      "loss": 1.0918,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.869330021743898e-05,
+      "loss": 0.9819,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8691759748046594e-05,
+      "loss": 1.08,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8690218434709355e-05,
+      "loss": 1.0383,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8688676277576916e-05,
+      "loss": 1.0693,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8687133276799027e-05,
+      "loss": 1.0205,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868558943252549e-05,
+      "loss": 0.968,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868404474490623e-05,
+      "loss": 1.1389,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868249921409122e-05,
+      "loss": 0.9498,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8680952840230526e-05,
+      "loss": 1.0131,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8679405623474294e-05,
+      "loss": 1.0602,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8677857563972766e-05,
+      "loss": 1.0781,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8676308661876242e-05,
+      "loss": 1.0237,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867475891733512e-05,
+      "loss": 1.1372,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8673208330499884e-05,
+      "loss": 1.027,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867165690152108e-05,
+      "loss": 1.116,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8670104630549348e-05,
+      "loss": 1.0724,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8668551517735417e-05,
+      "loss": 1.0875,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866699756323008e-05,
+      "loss": 1.0674,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8665442767184233e-05,
+      "loss": 1.0107,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866388712974883e-05,
+      "loss": 1.0161,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8662330651074926e-05,
+      "loss": 1.0909,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866077333131365e-05,
+      "loss": 1.0645,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8659215170616214e-05,
+      "loss": 1.0824,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8657656169133908e-05,
+      "loss": 1.0046,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865609632701811e-05,
+      "loss": 1.0159,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8654535644420276e-05,
+      "loss": 1.0816,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8652974121491936e-05,
+      "loss": 1.0458,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8651411758384718e-05,
+      "loss": 0.9793,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8649848555250324e-05,
+      "loss": 1.132,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8648284512240527e-05,
+      "loss": 1.0127,
+      "step": 2000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8646719629507203e-05,
+      "loss": 0.9684,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8645153907202285e-05,
+      "loss": 1.0386,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864358734547781e-05,
+      "loss": 1.0578,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8642019944485884e-05,
+      "loss": 1.0517,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.86404517043787e-05,
+      "loss": 1.064,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863888262530852e-05,
+      "loss": 0.9819,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8637312707427706e-05,
+      "loss": 0.9889,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863574195088869e-05,
+      "loss": 0.9502,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863417035584399e-05,
+      "loss": 1.0276,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8632597922446195e-05,
+      "loss": 1.0625,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8631024650847995e-05,
+      "loss": 1.0436,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8629450541202142e-05,
+      "loss": 1.0071,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8627875593661476e-05,
+      "loss": 0.9424,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8626299808378933e-05,
+      "loss": 1.0013,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8624723185507504e-05,
+      "loss": 1.0531,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862314572520028e-05,
+      "loss": 1.0145,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862156742761043e-05,
+      "loss": 1.0244,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861998829289119e-05,
+      "loss": 1.0334,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8618408321195906e-05,
+      "loss": 0.9859,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861682751267798e-05,
+      "loss": 1.0583,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8615245867490904e-05,
+      "loss": 1.0673,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861366338578825e-05,
+      "loss": 1.0352,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861208006772368e-05,
+      "loss": 1.0807,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8610495913450922e-05,
+      "loss": 1.0931,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8608910923123792e-05,
+      "loss": 1.1127,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8607325096896197e-05,
+      "loss": 1.0065,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860573843492211e-05,
+      "loss": 1.0729,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8604150937355588e-05,
+      "loss": 1.153,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8602562604350774e-05,
+      "loss": 0.9862,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.86009734360619e-05,
+      "loss": 1.0797,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8599383432643257e-05,
+      "loss": 0.9725,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8597792594249237e-05,
+      "loss": 1.0359,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.85962009210343e-05,
+      "loss": 1.0788,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8594608413153e-05,
+      "loss": 0.926,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8593015070759964e-05,
+      "loss": 1.0336,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8591420894009897e-05,
+      "loss": 1.0726,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8589825883057592e-05,
+      "loss": 1.0501,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8588230038057913e-05,
+      "loss": 1.047,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.858663335916582e-05,
+      "loss": 1.0063,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8585035846536347e-05,
+      "loss": 1.0552,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8583437500324605e-05,
+      "loss": 1.0493,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8581838320685782e-05,
+      "loss": 1.0632,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8580238307775164e-05,
+      "loss": 1.0989,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8578637461748105e-05,
+      "loss": 1.0125,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857703578276004e-05,
+      "loss": 1.1249,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857543327096649e-05,
+      "loss": 0.9747,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8573829926523047e-05,
+      "loss": 1.0419,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.85722257495854e-05,
+      "loss": 0.9659,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857062074030931e-05,
+      "loss": 1.0445,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.856901489885061e-05,
+      "loss": 1.1035,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8567408225365235e-05,
+      "loss": 1.0499,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.856580072000918e-05,
+      "loss": 0.9463,
+      "step": 2052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8564192382938526e-05,
+      "loss": 1.0505,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8562583214309447e-05,
+      "loss": 0.972,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8560973214278184e-05,
+      "loss": 1.0559,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855936238300106e-05,
+      "loss": 1.0287,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855775072063449e-05,
+      "loss": 1.0774,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8556138227334957e-05,
+      "loss": 1.1325,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855452490325903e-05,
+      "loss": 1.0364,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855291074856336e-05,
+      "loss": 1.1179,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8551295763404672e-05,
+      "loss": 1.0081,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8549679947939778e-05,
+      "loss": 1.017,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8548063302325573e-05,
+      "loss": 0.9778,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8546445826719023e-05,
+      "loss": 1.0427,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8544827521277187e-05,
+      "loss": 1.1354,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8543208386157195e-05,
+      "loss": 1.0229,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8541588421516255e-05,
+      "loss": 0.9796,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853996762751167e-05,
+      "loss": 1.0521,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8538346004300807e-05,
+      "loss": 1.068,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8536723552041124e-05,
+      "loss": 0.9638,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8535100270890157e-05,
+      "loss": 1.0831,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853347616100552e-05,
+      "loss": 1.0346,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853185122254491e-05,
+      "loss": 1.0084,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8530225455666103e-05,
+      "loss": 1.0645,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852859886052696e-05,
+      "loss": 0.9778,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8526971437285416e-05,
+      "loss": 1.0658,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852534318609949e-05,
+      "loss": 1.1572,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8523714107127278e-05,
+      "loss": 1.0576,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8522084200526962e-05,
+      "loss": 1.0164,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8520453466456797e-05,
+      "loss": 1.052,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851882190507513e-05,
+      "loss": 1.1089,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8517189516540376e-05,
+      "loss": 1.0274,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8515556301011036e-05,
+      "loss": 0.9376,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8513922258645687e-05,
+      "loss": 1.0569,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8512287389602996e-05,
+      "loss": 1.0268,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8510651694041702e-05,
+      "loss": 1.0364,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850901517212062e-05,
+      "loss": 0.9996,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8507377823998664e-05,
+      "loss": 1.0437,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8505739649834805e-05,
+      "loss": 0.9882,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850410064978811e-05,
+      "loss": 1.034,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8502460824017722e-05,
+      "loss": 0.9938,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8500820172682858e-05,
+      "loss": 1.0556,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8499178695942826e-05,
+      "loss": 1.0346,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8497536393957005e-05,
+      "loss": 1.0504,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.849589326688486e-05,
+      "loss": 1.057,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8494249314885932e-05,
+      "loss": 0.9636,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8492604538119846e-05,
+      "loss": 1.0475,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8490958936746304e-05,
+      "loss": 1.077,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8489312510925092e-05,
+      "loss": 1.173,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.848766526081607e-05,
+      "loss": 1.0449,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.848601718657918e-05,
+      "loss": 1.0365,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8484368288374452e-05,
+      "loss": 1.0242,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8482718566361978e-05,
+      "loss": 0.9973,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8481068020701954e-05,
+      "loss": 1.0527,
+      "step": 2104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.847941665155463e-05,
+      "loss": 1.0499,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8477764459080364e-05,
+      "loss": 0.9738,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8476111443439567e-05,
+      "loss": 1.0685,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8474457604792746e-05,
+      "loss": 1.0809,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8472802943300484e-05,
+      "loss": 1.0588,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8471147459123447e-05,
+      "loss": 1.0471,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8469491152422375e-05,
+      "loss": 1.1174,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8467834023358088e-05,
+      "loss": 1.0535,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8466176072091493e-05,
+      "loss": 1.0812,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846451729878357e-05,
+      "loss": 0.9984,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846285770359538e-05,
+      "loss": 1.141,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846119728668807e-05,
+      "loss": 1.0037,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8459536048222853e-05,
+      "loss": 1.0139,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.845787398836104e-05,
+      "loss": 1.0116,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8456211107264005e-05,
+      "loss": 1.1526,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8454547405093212e-05,
+      "loss": 0.9989,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8452882882010202e-05,
+      "loss": 0.9911,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8451217538176597e-05,
+      "loss": 1.0047,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8449551373754094e-05,
+      "loss": 1.0286,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.844788438890447e-05,
+      "loss": 0.9305,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8446216583789595e-05,
+      "loss": 1.13,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8444547958571396e-05,
+      "loss": 0.9819,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8442878513411896e-05,
+      "loss": 1.0258,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84412082484732e-05,
+      "loss": 1.0426,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8439537163917474e-05,
+      "loss": 1.019,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8437865259906987e-05,
+      "loss": 1.0707,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8436192536604066e-05,
+      "loss": 0.9826,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8434518994171136e-05,
+      "loss": 1.0392,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.843284463277069e-05,
+      "loss": 1.0442,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84311694525653e-05,
+      "loss": 1.069,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8429493453717624e-05,
+      "loss": 1.0777,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84278166363904e-05,
+      "loss": 0.985,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8426139000746436e-05,
+      "loss": 1.0585,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8424460546948632e-05,
+      "loss": 1.0593,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8422781275159955e-05,
+      "loss": 1.0302,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8421101185543463e-05,
+      "loss": 1.0857,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8419420278262283e-05,
+      "loss": 1.0411,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841773855347963e-05,
+      "loss": 0.9329,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8416056011358794e-05,
+      "loss": 1.0106,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841437265206314e-05,
+      "loss": 1.0563,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8412688475756124e-05,
+      "loss": 1.0244,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841100348260127e-05,
+      "loss": 1.0707,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840931767276219e-05,
+      "loss": 1.0178,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840763104640257e-05,
+      "loss": 0.9471,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8405943603686176e-05,
+      "loss": 1.0031,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8404255344776853e-05,
+      "loss": 0.9953,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8402566269838523e-05,
+      "loss": 1.0133,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84008763790352e-05,
+      "loss": 1.0757,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8399185672530957e-05,
+      "loss": 1.0574,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8397494150489963e-05,
+      "loss": 0.997,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.839580181307646e-05,
+      "loss": 1.0577,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8394108660454766e-05,
+      "loss": 1.0457,
+      "step": 2156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8392414692789283e-05,
+      "loss": 1.0812,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8390719910244487e-05,
+      "loss": 1.0812,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8389024312984944e-05,
+      "loss": 1.0905,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8387327901175286e-05,
+      "loss": 1.0247,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8385630674980228e-05,
+      "loss": 1.0009,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.838393263456457e-05,
+      "loss": 1.0431,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8382233780093186e-05,
+      "loss": 1.0324,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.838053411173103e-05,
+      "loss": 1.0877,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8378833629643135e-05,
+      "loss": 1.0442,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8377132333994606e-05,
+      "loss": 1.0651,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837543022495065e-05,
+      "loss": 1.1064,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837372730267652e-05,
+      "loss": 1.0363,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8372023567337575e-05,
+      "loss": 1.0295,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8370319019099236e-05,
+      "loss": 1.0425,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8368613658127013e-05,
+      "loss": 1.0125,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8366907484586497e-05,
+      "loss": 1.1053,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8365200498643344e-05,
+      "loss": 1.0566,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83634927004633e-05,
+      "loss": 1.0729,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.836178409021219e-05,
+      "loss": 1.0013,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8360074668055915e-05,
+      "loss": 1.0256,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.835836443416045e-05,
+      "loss": 1.0968,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8356653388691857e-05,
+      "loss": 1.0771,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8354941531816278e-05,
+      "loss": 0.9857,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8353228863699922e-05,
+      "loss": 1.0101,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8351515384509085e-05,
+      "loss": 1.0021,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8349801094410148e-05,
+      "loss": 1.0647,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8348085993569556e-05,
+      "loss": 1.0537,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8346370082153843e-05,
+      "loss": 1.0185,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.834465336032962e-05,
+      "loss": 1.0331,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8342935828263574e-05,
+      "loss": 1.1253,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8341217486122474e-05,
+      "loss": 1.0611,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8339498334073166e-05,
+      "loss": 0.9664,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8337778372282577e-05,
+      "loss": 1.0025,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83360576009177e-05,
+      "loss": 1.0703,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8334336020145625e-05,
+      "loss": 1.1803,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833261363013352e-05,
+      "loss": 1.0838,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833089043104861e-05,
+      "loss": 1.0218,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832916642305822e-05,
+      "loss": 1.0724,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8327441606329747e-05,
+      "loss": 1.0533,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832571598103066e-05,
+      "loss": 1.1072,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8323989547328515e-05,
+      "loss": 1.1078,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8322262305390948e-05,
+      "loss": 1.0436,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8320534255385663e-05,
+      "loss": 0.9866,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8318805397480455e-05,
+      "loss": 0.992,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8317075731843185e-05,
+      "loss": 1.1087,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8315345258641802e-05,
+      "loss": 1.0245,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8313613978044332e-05,
+      "loss": 0.9813,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8311881890218873e-05,
+      "loss": 1.0345,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8310148995333604e-05,
+      "loss": 0.978,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.830841529355679e-05,
+      "loss": 1.1189,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8306680785056767e-05,
+      "loss": 1.0391,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8304945470001948e-05,
+      "loss": 1.0214,
+      "step": 2208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.830320934856083e-05,
+      "loss": 0.9973,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8301472420901985e-05,
+      "loss": 0.9503,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829973468719406e-05,
+      "loss": 0.933,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8297996147605787e-05,
+      "loss": 1.1116,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8296256802305975e-05,
+      "loss": 1.0808,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829451665146351e-05,
+      "loss": 1.0453,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8292775695247344e-05,
+      "loss": 1.0282,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8291033933826535e-05,
+      "loss": 0.9474,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8289291367370193e-05,
+      "loss": 0.9736,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8287547996047523e-05,
+      "loss": 1.0175,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.828580382002779e-05,
+      "loss": 1.0677,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8284058839480363e-05,
+      "loss": 1.0451,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8282313054574667e-05,
+      "loss": 1.1505,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8280566465480206e-05,
+      "loss": 1.0068,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8278819072366586e-05,
+      "loss": 1.0407,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8277070875403455e-05,
+      "loss": 1.0591,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8275321874760573e-05,
+      "loss": 0.9545,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8273572070607756e-05,
+      "loss": 1.127,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.82718214631149e-05,
+      "loss": 1.0239,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8270070052451995e-05,
+      "loss": 0.9695,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826831783878909e-05,
+      "loss": 0.9777,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8266564822296323e-05,
+      "loss": 1.0942,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8264811003143907e-05,
+      "loss": 0.9792,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826305638150213e-05,
+      "loss": 0.9391,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8261300957541362e-05,
+      "loss": 1.038,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.825954473143205e-05,
+      "loss": 1.0258,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8257787703344722e-05,
+      "loss": 1.0996,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8256029873449976e-05,
+      "loss": 1.0936,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8254271241918495e-05,
+      "loss": 1.0576,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.825251180892103e-05,
+      "loss": 1.0597,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.825075157462843e-05,
+      "loss": 1.0363,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8248990539211596e-05,
+      "loss": 1.0679,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8247228702841523e-05,
+      "loss": 1.0161,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8245466065689282e-05,
+      "loss": 1.0477,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8243702627926027e-05,
+      "loss": 1.0283,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.824193838972297e-05,
+      "loss": 1.0766,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8240173351251415e-05,
+      "loss": 1.0016,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823840751268275e-05,
+      "loss": 1.0085,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823664087418843e-05,
+      "loss": 0.9621,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8234873435939987e-05,
+      "loss": 1.1148,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823310519810904e-05,
+      "loss": 1.021,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8231336160867275e-05,
+      "loss": 0.9477,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8229566324386462e-05,
+      "loss": 0.997,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8227795688838446e-05,
+      "loss": 0.9951,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822602425439515e-05,
+      "loss": 1.0912,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822425202122858e-05,
+      "loss": 1.0586,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8222478989510808e-05,
+      "loss": 1.1003,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8220705159413996e-05,
+      "loss": 1.0667,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821893053111037e-05,
+      "loss": 1.0029,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8217155104772256e-05,
+      "loss": 1.0842,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8215378880572028e-05,
+      "loss": 0.9427,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8213601858682158e-05,
+      "loss": 1.049,
+      "step": 2260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821182403927519e-05,
+      "loss": 0.8957,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8210045422523744e-05,
+      "loss": 1.0703,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8208266008600522e-05,
+      "loss": 1.0033,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8206485797678294e-05,
+      "loss": 0.9969,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820470478992992e-05,
+      "loss": 0.9482,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820292298552832e-05,
+      "loss": 1.0342,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8201140384646518e-05,
+      "loss": 1.1195,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819935698745759e-05,
+      "loss": 1.0099,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8197572794134693e-05,
+      "loss": 1.0109,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8195787804851076e-05,
+      "loss": 1.055,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8194002019780053e-05,
+      "loss": 1.0043,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8192215439095025e-05,
+      "loss": 0.9746,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8190428062969454e-05,
+      "loss": 1.0851,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8188639891576893e-05,
+      "loss": 1.1168,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818685092509097e-05,
+      "loss": 1.0485,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8185061163685386e-05,
+      "loss": 1.0419,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818327060753392e-05,
+      "loss": 1.0283,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818147925681044e-05,
+      "loss": 1.0092,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.817968711168887e-05,
+      "loss": 1.0701,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8177894172343227e-05,
+      "loss": 0.9814,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8176100438947598e-05,
+      "loss": 1.0785,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.817430591167615e-05,
+      "loss": 0.9996,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8172510590703125e-05,
+      "loss": 1.064,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8170714476202848e-05,
+      "loss": 0.9909,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8168917568349714e-05,
+      "loss": 1.0093,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8167119867318197e-05,
+      "loss": 1.1086,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816532137328285e-05,
+      "loss": 1.037,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.81635220864183e-05,
+      "loss": 1.1113,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816172200689925e-05,
+      "loss": 1.0875,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8159921134900486e-05,
+      "loss": 1.0753,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8158119470596872e-05,
+      "loss": 0.991,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8156317014163337e-05,
+      "loss": 1.0705,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8154513765774897e-05,
+      "loss": 1.0127,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8152709725606642e-05,
+      "loss": 1.0273,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815090489383374e-05,
+      "loss": 1.0685,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8149099270631434e-05,
+      "loss": 1.0586,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8147292856175048e-05,
+      "loss": 1.0211,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8145485650639973e-05,
+      "loss": 1.0395,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8143677654201686e-05,
+      "loss": 1.0173,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8141868867035745e-05,
+      "loss": 1.014,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8140059289317768e-05,
+      "loss": 1.0308,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8138248921223465e-05,
+      "loss": 1.093,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813643776292862e-05,
+      "loss": 1.025,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8134625814609084e-05,
+      "loss": 1.0765,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8132813076440796e-05,
+      "loss": 1.0483,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8130999548599767e-05,
+      "loss": 1.1011,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.812918523126209e-05,
+      "loss": 1.1081,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8127370124603927e-05,
+      "loss": 0.9886,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8125554228801518e-05,
+      "loss": 1.0382,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8123737544031178e-05,
+      "loss": 1.0908,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8121920070469307e-05,
+      "loss": 1.1217,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8120101808292373e-05,
+      "loss": 1.0139,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811828275767693e-05,
+      "loss": 1.0166,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81164629187996e-05,
+      "loss": 0.9346,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811464229183708e-05,
+      "loss": 1.0475,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811282087696615e-05,
+      "loss": 1.152,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811099867436367e-05,
+      "loss": 1.1,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8109175684206558e-05,
+      "loss": 1.0071,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8107351906671834e-05,
+      "loss": 1.0507,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8105527341936574e-05,
+      "loss": 1.1061,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.810370199017794e-05,
+      "loss": 0.9915,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.810187585157317e-05,
+      "loss": 1.0325,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8100048926299575e-05,
+      "loss": 1.0447,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8098221214534543e-05,
+      "loss": 1.0671,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809639271645554e-05,
+      "loss": 1.0481,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8094563432240107e-05,
+      "loss": 1.0434,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8092733362065867e-05,
+      "loss": 0.9598,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8090902506110513e-05,
+      "loss": 1.0388,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8089070864551815e-05,
+      "loss": 1.0159,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8087238437567614e-05,
+      "loss": 0.92,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8085405225335845e-05,
+      "loss": 1.0074,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8083571228034498e-05,
+      "loss": 1.0735,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8081736445841652e-05,
+      "loss": 1.0328,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807990087893546e-05,
+      "loss": 1.0824,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8078064527494153e-05,
+      "loss": 1.0205,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807622739169603e-05,
+      "loss": 1.0946,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8074389471719473e-05,
+      "loss": 1.0281,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807255076774294e-05,
+      "loss": 1.0146,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8070711279944965e-05,
+      "loss": 1.0603,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8068871008504153e-05,
+      "loss": 1.0092,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8067029953599193e-05,
+      "loss": 1.0954,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8065188115408844e-05,
+      "loss": 1.077,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8063345494111947e-05,
+      "loss": 1.0367,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8061502089887406e-05,
+      "loss": 1.0577,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805965790291422e-05,
+      "loss": 1.09,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805781293337145e-05,
+      "loss": 1.0407,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8055967181438236e-05,
+      "loss": 1.1013,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8054120647293798e-05,
+      "loss": 1.0609,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805227333111743e-05,
+      "loss": 1.1115,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8050425233088496e-05,
+      "loss": 1.007,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8048576353386448e-05,
+      "loss": 1.0793,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80467266921908e-05,
+      "loss": 1.0655,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.804487624968115e-05,
+      "loss": 1.0771,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8043025026037178e-05,
+      "loss": 1.0975,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8041173021438623e-05,
+      "loss": 1.0847,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8039320236065314e-05,
+      "loss": 1.0543,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803746667009715e-05,
+      "loss": 1.0979,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803561232371411e-05,
+      "loss": 1.0743,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803375719709624e-05,
+      "loss": 0.9973,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803190129042367e-05,
+      "loss": 1.0715,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8030044603876607e-05,
+      "loss": 1.0442,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8028187137635325e-05,
+      "loss": 0.9943,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.802632889188018e-05,
+      "loss": 1.0232,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8024469866791602e-05,
+      "loss": 1.0443,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8022610062550097e-05,
+      "loss": 1.0583,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.802074947933625e-05,
+      "loss": 1.1072,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8018888117330714e-05,
+      "loss": 1.0684,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.801702597671422e-05,
+      "loss": 1.0039,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8015163057667584e-05,
+      "loss": 1.0413,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8013299360371685e-05,
+      "loss": 1.1208,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8011434885007482e-05,
+      "loss": 1.1572,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8009569631756013e-05,
+      "loss": 1.1047,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800770360079839e-05,
+      "loss": 1.0545,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8005836792315793e-05,
+      "loss": 1.0554,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8003969206489486e-05,
+      "loss": 0.9958,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800210084350081e-05,
+      "loss": 1.041,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8000231703531176e-05,
+      "loss": 1.0764,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.799836178676207e-05,
+      "loss": 1.0009,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.799649109337506e-05,
+      "loss": 1.0854,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.799461962355178e-05,
+      "loss": 1.0034,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7992747377473945e-05,
+      "loss": 1.066,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7990874355323345e-05,
+      "loss": 0.9973,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7989000557281848e-05,
+      "loss": 1.0406,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7987125983531393e-05,
+      "loss": 1.048,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7985250634253994e-05,
+      "loss": 1.0377,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7983374509631742e-05,
+      "loss": 0.9792,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7981497609846804e-05,
+      "loss": 1.0044,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7979619935081424e-05,
+      "loss": 1.0297,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7977741485517918e-05,
+      "loss": 0.9548,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797586226133867e-05,
+      "loss": 1.0643,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7973982262726163e-05,
+      "loss": 1.0259,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7972101489862924e-05,
+      "loss": 1.0184,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7970219942931583e-05,
+      "loss": 0.9373,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7968337622114824e-05,
+      "loss": 1.0118,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796645452759542e-05,
+      "loss": 1.0542,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7964570659556206e-05,
+      "loss": 1.0573,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796268601818011e-05,
+      "loss": 1.0338,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796080060365012e-05,
+      "loss": 1.0849,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.795891441614931e-05,
+      "loss": 0.9721,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7957027455860815e-05,
+      "loss": 0.9661,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7955139722967856e-05,
+      "loss": 1.0515,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.795325121765373e-05,
+      "loss": 1.0646,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7951361940101806e-05,
+      "loss": 1.0642,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.794947189049552e-05,
+      "loss": 1.0267,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7947581069018398e-05,
+      "loss": 1.0552,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7945689475854033e-05,
+      "loss": 1.043,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.794379711118609e-05,
+      "loss": 1.072,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7941903975198305e-05,
+      "loss": 1.0809,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7940010068074508e-05,
+      "loss": 1.011,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7938115389998595e-05,
+      "loss": 1.0013,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7936219941154518e-05,
+      "loss": 1.075,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7934323721726334e-05,
+      "loss": 0.9973,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793242673189815e-05,
+      "loss": 1.0875,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7930528971854166e-05,
+      "loss": 1.0084,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.792863044177864e-05,
+      "loss": 0.9905,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.792673114185593e-05,
+      "loss": 1.0517,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7924831072270436e-05,
+      "loss": 0.994,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7922930233206656e-05,
+      "loss": 1.0656,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7921028624849158e-05,
+      "loss": 1.0317,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7919126247382576e-05,
+      "loss": 1.0267,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7917223100991635e-05,
+      "loss": 1.0253,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791531918586112e-05,
+      "loss": 1.028,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791341450217589e-05,
+      "loss": 1.0748,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7911509050120892e-05,
+      "loss": 1.0347,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7909602829881137e-05,
+      "loss": 0.9846,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7907695841641716e-05,
+      "loss": 0.962,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790578808558779e-05,
+      "loss": 0.9655,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7903879561904597e-05,
+      "loss": 1.0653,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7901970270777446e-05,
+      "loss": 1.1123,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790006021239173e-05,
+      "loss": 0.9634,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7898149386932905e-05,
+      "loss": 1.0519,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789623779458651e-05,
+      "loss": 1.0199,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7894325435538156e-05,
+      "loss": 1.0568,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789241230997352e-05,
+      "loss": 1.0159,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789049841807837e-05,
+      "loss": 1.0239,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7888583760038534e-05,
+      "loss": 1.0311,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.788666833603992e-05,
+      "loss": 1.009,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7884752146268513e-05,
+      "loss": 1.0779,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7882835190910367e-05,
+      "loss": 1.0826,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7880917470151614e-05,
+      "loss": 0.9293,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787899898417846e-05,
+      "loss": 0.9554,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7877079733177185e-05,
+      "loss": 0.9906,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787515971733414e-05,
+      "loss": 1.0569,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7873238936835754e-05,
+      "loss": 0.9851,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7871317391868527e-05,
+      "loss": 1.0034,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.786939508261904e-05,
+      "loss": 1.0684,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.786747200927394e-05,
+      "loss": 1.0627,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.786554817201996e-05,
+      "loss": 0.9724,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7863623571043886e-05,
+      "loss": 1.0774,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.78616982065326e-05,
+      "loss": 1.042,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785977207867305e-05,
+      "loss": 1.0121,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785784518765225e-05,
+      "loss": 0.9614,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7855917533657302e-05,
+      "loss": 1.0562,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7853989116875373e-05,
+      "loss": 1.031,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7852059937493707e-05,
+      "loss": 0.9858,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7850129995699626e-05,
+      "loss": 1.0185,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7848199291680515e-05,
+      "loss": 1.0481,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7846267825623843e-05,
+      "loss": 0.9801,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.784433559771715e-05,
+      "loss": 0.9844,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7842402608148053e-05,
+      "loss": 0.9489,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7840468857104236e-05,
+      "loss": 1.088,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7838534344773453e-05,
+      "loss": 1.0467,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7836599071343558e-05,
+      "loss": 1.0226,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7834663037002444e-05,
+      "loss": 1.0098,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.78327262419381e-05,
+      "loss": 0.999,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7830788686338586e-05,
+      "loss": 1.0396,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.782885037039203e-05,
+      "loss": 1.0635,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7826911294286636e-05,
+      "loss": 1.0645,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7824971458210686e-05,
+      "loss": 1.1306,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.782303086235253e-05,
+      "loss": 1.0033,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7821089506900594e-05,
+      "loss": 1.0637,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.781914739204338e-05,
+      "loss": 1.0538,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7817204517969464e-05,
+      "loss": 1.016,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7815260884867486e-05,
+      "loss": 1.0062,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7813316492926173e-05,
+      "loss": 1.0084,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.781137134233432e-05,
+      "loss": 1.0087,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.780942543328079e-05,
+      "loss": 1.0704,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7807478765954532e-05,
+      "loss": 1.0439,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7805531340544554e-05,
+      "loss": 0.9776,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7803583157239958e-05,
+      "loss": 1.0364,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7801634216229892e-05,
+      "loss": 1.1217,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7799684517703605e-05,
+      "loss": 1.0322,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7797734061850397e-05,
+      "loss": 1.0562,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779578284885966e-05,
+      "loss": 1.0559,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779383087892085e-05,
+      "loss": 1.0043,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779187815222349e-05,
+      "loss": 1.0143,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7789924668957192e-05,
+      "loss": 1.0973,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778797042931163e-05,
+      "loss": 1.0002,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778601543347656e-05,
+      "loss": 1.0134,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7784059681641798e-05,
+      "loss": 0.9424,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778210317399725e-05,
+      "loss": 1.0309,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778014591073288e-05,
+      "loss": 1.0291,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7778187892038738e-05,
+      "loss": 0.9886,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777622911810494e-05,
+      "loss": 1.0247,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777426958912168e-05,
+      "loss": 1.1365,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777230930527922e-05,
+      "loss": 1.0692,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7770348266767897e-05,
+      "loss": 1.0869,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7768386473778124e-05,
+      "loss": 1.0273,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7766423926500382e-05,
+      "loss": 0.9996,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7764460625125236e-05,
+      "loss": 0.9023,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.776249656984331e-05,
+      "loss": 1.0679,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.776053176084531e-05,
+      "loss": 0.9586,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7758566198322014e-05,
+      "loss": 1.0402,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7756599882464274e-05,
+      "loss": 1.0235,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775463281346301e-05,
+      "loss": 1.0088,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7752664991509224e-05,
+      "loss": 1.0162,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775069641679398e-05,
+      "loss": 1.0695,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7748727089508423e-05,
+      "loss": 1.0068,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.774675700984377e-05,
+      "loss": 0.9468,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7744786177991307e-05,
+      "loss": 1.0903,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7742814594142405e-05,
+      "loss": 0.999,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.774084225848849e-05,
+      "loss": 0.988,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.773886917122107e-05,
+      "loss": 0.9908,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.773689533253173e-05,
+      "loss": 0.9036,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7734920742612125e-05,
+      "loss": 1.0569,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7732945401653978e-05,
+      "loss": 1.0361,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7730969309849093e-05,
+      "loss": 1.024,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7728992467389342e-05,
+      "loss": 1.0257,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7727014874466664e-05,
+      "loss": 0.9449,
+      "step": 2519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7725036531273087e-05,
+      "loss": 1.0485,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.77230574380007e-05,
+      "loss": 1.1176,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7721077594841663e-05,
+      "loss": 1.0071,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7719097001988216e-05,
+      "loss": 1.0302,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.771711565963267e-05,
+      "loss": 1.027,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7715133567967405e-05,
+      "loss": 1.1689,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7713150727184878e-05,
+      "loss": 1.0616,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7711167137477615e-05,
+      "loss": 1.0856,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.770918279903822e-05,
+      "loss": 1.0641,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7707197712059366e-05,
+      "loss": 1.0333,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.77052118767338e-05,
+      "loss": 1.0449,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.770322529325434e-05,
+      "loss": 0.9647,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7701237961813874e-05,
+      "loss": 1.0027,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7699249882605372e-05,
+      "loss": 1.057,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7697261055821864e-05,
+      "loss": 0.9373,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7695271481656467e-05,
+      "loss": 0.9232,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7693281160302354e-05,
+      "loss": 1.0239,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769129009195279e-05,
+      "loss": 1.0629,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7689298276801095e-05,
+      "loss": 1.0293,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.768730571504067e-05,
+      "loss": 1.0561,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7685312406864986e-05,
+      "loss": 1.0044,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.768331835246759e-05,
+      "loss": 0.9659,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7681323552042094e-05,
+      "loss": 1.0435,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7679328005782194e-05,
+      "loss": 1.0525,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767733171388165e-05,
+      "loss": 0.9157,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767533467653429e-05,
+      "loss": 1.0287,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7673336893934033e-05,
+      "loss": 0.9607,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7671338366274846e-05,
+      "loss": 1.0001,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7669339093750786e-05,
+      "loss": 1.0719,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7667339076555972e-05,
+      "loss": 1.0424,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766533831488461e-05,
+      "loss": 0.9742,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7663336808930958e-05,
+      "loss": 0.9994,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7661334558889357e-05,
+      "loss": 0.8878,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7659331564954225e-05,
+      "loss": 1.0504,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7657327827320046e-05,
+      "loss": 1.0505,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7655323346181376e-05,
+      "loss": 1.0079,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.765331812173284e-05,
+      "loss": 0.9163,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7651312154169145e-05,
+      "loss": 1.0359,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7649305443685068e-05,
+      "loss": 1.0269,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7647297990475447e-05,
+      "loss": 1.0901,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.76452897947352e-05,
+      "loss": 0.9765,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7643280856659324e-05,
+      "loss": 1.0106,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7641271176442876e-05,
+      "loss": 0.9799,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7639260754280993e-05,
+      "loss": 1.0577,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7637249590368878e-05,
+      "loss": 1.1433,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763523768490181e-05,
+      "loss": 0.942,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763322503807514e-05,
+      "loss": 0.9434,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7631211650084294e-05,
+      "loss": 0.9318,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7629197521124758e-05,
+      "loss": 1.0207,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7627182651392105e-05,
+      "loss": 1.0662,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7625167041081967e-05,
+      "loss": 1.0596,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.762315069039006e-05,
+      "loss": 1.0049,
+      "step": 2571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7621133599512163e-05,
+      "loss": 0.9696,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.761911576864413e-05,
+      "loss": 1.0704,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.761709719798189e-05,
+      "loss": 1.0134,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7615077887721432e-05,
+      "loss": 1.019,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.761305783805883e-05,
+      "loss": 1.0597,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.761103704919023e-05,
+      "loss": 1.0136,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7609015521311836e-05,
+      "loss": 1.0345,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.760699325461994e-05,
+      "loss": 0.979,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7604970249310893e-05,
+      "loss": 1.0294,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7602946505581128e-05,
+      "loss": 1.1122,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7600922023627137e-05,
+      "loss": 0.9683,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7598896803645498e-05,
+      "loss": 1.0065,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.759687084583285e-05,
+      "loss": 1.0426,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7594844150385912e-05,
+      "loss": 1.0529,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.759281671750147e-05,
+      "loss": 1.0054,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7590788547376375e-05,
+      "loss": 1.0537,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7588759640207564e-05,
+      "loss": 1.1071,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7586729996192032e-05,
+      "loss": 0.9615,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7584699615526857e-05,
+      "loss": 1.0347,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7582668498409185e-05,
+      "loss": 1.0728,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7580636645036224e-05,
+      "loss": 1.0602,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7578604055605266e-05,
+      "loss": 1.0139,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757657073031367e-05,
+      "loss": 1.0336,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757453666935886e-05,
+      "loss": 1.0527,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7572501872938343e-05,
+      "loss": 1.0498,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7570466341249694e-05,
+      "loss": 1.0136,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.756843007449055e-05,
+      "loss": 1.0179,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7566393072858633e-05,
+      "loss": 1.0139,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7564355336551727e-05,
+      "loss": 0.9813,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.756231686576769e-05,
+      "loss": 1.0475,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7560277660704455e-05,
+      "loss": 1.0939,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7558237721560017e-05,
+      "loss": 0.9999,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.755619704853246e-05,
+      "loss": 0.9882,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7554155641819913e-05,
+      "loss": 1.0436,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7552113501620595e-05,
+      "loss": 1.0463,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7550070628132797e-05,
+      "loss": 1.088,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7548027021554874e-05,
+      "loss": 1.0583,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754598268208525e-05,
+      "loss": 0.9847,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754393760992243e-05,
+      "loss": 1.0344,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7541891805264984e-05,
+      "loss": 1.0948,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7539845268311548e-05,
+      "loss": 1.0298,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7537797999260843e-05,
+      "loss": 1.0408,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7535749998311645e-05,
+      "loss": 0.9698,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7533701265662818e-05,
+      "loss": 1.0325,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.753165180151328e-05,
+      "loss": 1.0237,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7529601606062033e-05,
+      "loss": 1.0992,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752755067950814e-05,
+      "loss": 1.0771,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752549902205075e-05,
+      "loss": 1.0891,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752344663388906e-05,
+      "loss": 1.1505,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7521393515222358e-05,
+      "loss": 1.0652,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7519339666249997e-05,
+      "loss": 0.967,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7517285087171398e-05,
+      "loss": 1.035,
+      "step": 2623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7515229778186052e-05,
+      "loss": 1.0078,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.751317373949353e-05,
+      "loss": 1.113,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7511116971293463e-05,
+      "loss": 1.0397,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.750905947378555e-05,
+      "loss": 1.0714,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7507001247169587e-05,
+      "loss": 0.9449,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7504942291645408e-05,
+      "loss": 1.0286,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7502882607412933e-05,
+      "loss": 1.0053,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7500822194672153e-05,
+      "loss": 1.0352,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749876105362313e-05,
+      "loss": 0.9661,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7496699184465994e-05,
+      "loss": 0.9808,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7494636587400942e-05,
+      "loss": 1.0523,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7492573262628253e-05,
+      "loss": 0.9927,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749050921034826e-05,
+      "loss": 0.997,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7488444430761394e-05,
+      "loss": 1.1086,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7486378924068123e-05,
+      "loss": 0.9932,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7484312690469004e-05,
+      "loss": 1.0398,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.748224573016467e-05,
+      "loss": 1.0585,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7480178043355814e-05,
+      "loss": 1.0385,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7478109630243195e-05,
+      "loss": 1.0131,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747604049102766e-05,
+      "loss": 1.0638,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747397062591011e-05,
+      "loss": 1.0418,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747190003509153e-05,
+      "loss": 0.9743,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.746982871877296e-05,
+      "loss": 1.0271,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.746775667715552e-05,
+      "loss": 0.9581,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7465683910440405e-05,
+      "loss": 0.9884,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7463610418828872e-05,
+      "loss": 1.0553,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7461536202522248e-05,
+      "loss": 1.0446,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.745946126172194e-05,
+      "loss": 1.0236,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.745738559662941e-05,
+      "loss": 1.0433,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7455309207446206e-05,
+      "loss": 1.0686,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7453232094373936e-05,
+      "loss": 1.0065,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7451154257614287e-05,
+      "loss": 1.1107,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7449075697369005e-05,
+      "loss": 1.0189,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.744699641383991e-05,
+      "loss": 1.001,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7444916407228904e-05,
+      "loss": 0.9569,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.744283567773794e-05,
+      "loss": 1.0637,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.744075422556906e-05,
+      "loss": 0.985,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743867205092436e-05,
+      "loss": 1.0247,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7436589154006014e-05,
+      "loss": 0.9716,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7434505535016266e-05,
+      "loss": 0.9887,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743242119415743e-05,
+      "loss": 1.0084,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7430336131631895e-05,
+      "loss": 0.9658,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7428250347642102e-05,
+      "loss": 1.0571,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7426163842390585e-05,
+      "loss": 1.0304,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7424076616079933e-05,
+      "loss": 1.0391,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7421988668912812e-05,
+      "loss": 1.0172,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7419900001091953e-05,
+      "loss": 1.0193,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7417810612820164e-05,
+      "loss": 1.0075,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7415720504300314e-05,
+      "loss": 0.9739,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.741362967573535e-05,
+      "loss": 0.9548,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.741153812732828e-05,
+      "loss": 1.0187,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.74094458592822e-05,
+      "loss": 1.0002,
+      "step": 2675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7407352871800246e-05,
+      "loss": 1.0058,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.740525916508565e-05,
+      "loss": 0.9621,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7403164739341708e-05,
+      "loss": 0.9763,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7401069594771777e-05,
+      "loss": 0.9875,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.739897373157929e-05,
+      "loss": 1.0526,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7396877149967752e-05,
+      "loss": 1.0228,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7394779850140736e-05,
+      "loss": 1.0105,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7392681832301877e-05,
+      "loss": 1.0886,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7390583096654895e-05,
+      "loss": 1.0549,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7388483643403565e-05,
+      "loss": 1.0441,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7386383472751745e-05,
+      "loss": 1.0606,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7384282584903345e-05,
+      "loss": 1.0312,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7382180980062365e-05,
+      "loss": 1.0511,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.738007865843286e-05,
+      "loss": 1.0842,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7377975620218954e-05,
+      "loss": 1.0897,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7375871865624858e-05,
+      "loss": 0.9657,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7373767394854836e-05,
+      "loss": 1.0941,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737166220811322e-05,
+      "loss": 1.0516,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7369556305604422e-05,
+      "loss": 1.0109,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7367449687532923e-05,
+      "loss": 1.0388,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736534235410326e-05,
+      "loss": 1.0449,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736323430552006e-05,
+      "loss": 0.9805,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7361125541988e-05,
+      "loss": 1.0132,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.735901606371184e-05,
+      "loss": 0.9883,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7356905870896407e-05,
+      "loss": 1.024,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.735479496374658e-05,
+      "loss": 1.0895,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.735268334246734e-05,
+      "loss": 1.0265,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7350571007263708e-05,
+      "loss": 0.9832,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7348457958340792e-05,
+      "loss": 0.9358,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7346344195903756e-05,
+      "loss": 0.9479,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7344229720157846e-05,
+      "loss": 1.0958,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734211453130837e-05,
+      "loss": 0.9758,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7339998629560705e-05,
+      "loss": 0.9993,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7337882015120303e-05,
+      "loss": 1.0138,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7335764688192676e-05,
+      "loss": 1.0307,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7333646648983416e-05,
+      "loss": 0.9885,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733152789769817e-05,
+      "loss": 0.9667,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7329408434542673e-05,
+      "loss": 1.0549,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7327288259722714e-05,
+      "loss": 1.0446,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7325167373444153e-05,
+      "loss": 1.0512,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7323045775912927e-05,
+      "loss": 1.0763,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7320923467335035e-05,
+      "loss": 0.9591,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7318800447916543e-05,
+      "loss": 1.0386,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7316676717863598e-05,
+      "loss": 0.959,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7314552277382403e-05,
+      "loss": 1.0319,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731242712667924e-05,
+      "loss": 1.0431,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7310301265960446e-05,
+      "loss": 1.0417,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730817469543244e-05,
+      "loss": 1.0317,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7306047415301706e-05,
+      "loss": 0.9843,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7303919425774803e-05,
+      "loss": 1.0439,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7301790727058344e-05,
+      "loss": 1.0338,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7299661319359024e-05,
+      "loss": 0.9682,
+      "step": 2727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7297531202883598e-05,
+      "loss": 1.0064,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.72954003778389e-05,
+      "loss": 1.063,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7293268844431826e-05,
+      "loss": 1.0063,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7291136602869337e-05,
+      "loss": 1.064,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7289003653358472e-05,
+      "loss": 0.9403,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7286869996106335e-05,
+      "loss": 0.9747,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7284735631320093e-05,
+      "loss": 1.0442,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.728260055920699e-05,
+      "loss": 1.087,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7280464779974335e-05,
+      "loss": 1.0121,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7278328293829506e-05,
+      "loss": 1.0092,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7276191100979952e-05,
+      "loss": 1.019,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727405320163318e-05,
+      "loss": 1.0717,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7271914595996784e-05,
+      "loss": 1.0173,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7269775284278413e-05,
+      "loss": 1.0306,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7267635266685782e-05,
+      "loss": 1.0286,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7265494543426692e-05,
+      "loss": 1.0643,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7263353114708993e-05,
+      "loss": 0.9961,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.726121098074061e-05,
+      "loss": 0.9495,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7259068141729542e-05,
+      "loss": 1.0167,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7256924597883854e-05,
+      "loss": 1.0314,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7254780349411677e-05,
+      "loss": 1.0047,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.725263539652121e-05,
+      "loss": 1.0306,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7250489739420718e-05,
+      "loss": 1.0633,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7248343378318545e-05,
+      "loss": 1.0102,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7246196313423095e-05,
+      "loss": 0.9808,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.724404854494284e-05,
+      "loss": 1.0411,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7241900073086318e-05,
+      "loss": 1.0381,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7239750898062146e-05,
+      "loss": 0.9133,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7237601020079003e-05,
+      "loss": 1.0281,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.723545043934563e-05,
+      "loss": 0.9606,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7233299156070852e-05,
+      "loss": 1.0821,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7231147170463538e-05,
+      "loss": 1.0383,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7228994482732653e-05,
+      "loss": 1.0551,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.722684109308721e-05,
+      "loss": 1.0386,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.72246870017363e-05,
+      "loss": 1.0535,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7222532208889076e-05,
+      "loss": 1.0576,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7220376714754766e-05,
+      "loss": 1.0323,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.721822051954266e-05,
+      "loss": 0.9969,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7216063623462112e-05,
+      "loss": 1.1132,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7213906026722566e-05,
+      "loss": 1.0056,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7211747729533504e-05,
+      "loss": 1.0266,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7209588732104497e-05,
+      "loss": 1.0036,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7207429034645176e-05,
+      "loss": 1.0745,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7205268637365244e-05,
+      "loss": 1.0687,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720310754047446e-05,
+      "loss": 1.1022,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7200945744182673e-05,
+      "loss": 1.0574,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719878324869978e-05,
+      "loss": 0.9339,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7196620054235755e-05,
+      "loss": 1.1259,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7194456161000634e-05,
+      "loss": 1.0225,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7192291569204535e-05,
+      "loss": 0.9438,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719012627905762e-05,
+      "loss": 1.0238,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7187960290770144e-05,
+      "loss": 1.0552,
+      "step": 2779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.718579360455241e-05,
+      "loss": 1.0419,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7183626220614804e-05,
+      "loss": 1.0032,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7181458139167767e-05,
+      "loss": 0.9857,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717928936042182e-05,
+      "loss": 0.9947,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7177119884587536e-05,
+      "loss": 1.0049,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717494971187557e-05,
+      "loss": 1.0354,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717277884249664e-05,
+      "loss": 1.0414,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717060727666153e-05,
+      "loss": 0.9809,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.716843501458109e-05,
+      "loss": 0.9442,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7166262056466253e-05,
+      "loss": 1.0669,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.716408840252799e-05,
+      "loss": 1.0515,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7161914052977368e-05,
+      "loss": 0.9689,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7159739008025503e-05,
+      "loss": 1.0187,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7157563267883594e-05,
+      "loss": 1.0905,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7155386832762892e-05,
+      "loss": 1.0617,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.715320970287472e-05,
+      "loss": 0.9612,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.715103187843048e-05,
+      "loss": 1.0009,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7148853359641627e-05,
+      "loss": 1.0511,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7146674146719688e-05,
+      "loss": 1.0804,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7144494239876263e-05,
+      "loss": 1.114,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7142313639323012e-05,
+      "loss": 1.0361,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7140132345271662e-05,
+      "loss": 1.1031,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7137950357934017e-05,
+      "loss": 0.9792,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7135767677521938e-05,
+      "loss": 0.9985,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7133584304247354e-05,
+      "loss": 1.0551,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7131400238322267e-05,
+      "loss": 1.0125,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7129215479958747e-05,
+      "loss": 0.9788,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7127030029368925e-05,
+      "loss": 0.9541,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7124843886765e-05,
+      "loss": 0.9254,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712265705235924e-05,
+      "loss": 1.0075,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712046952636398e-05,
+      "loss": 1.0048,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7118281308991632e-05,
+      "loss": 0.9783,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7116092400454655e-05,
+      "loss": 1.0132,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7113902800965588e-05,
+      "loss": 1.0033,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7111712510737035e-05,
+      "loss": 1.0657,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.710952152998167e-05,
+      "loss": 1.1212,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7107329858912226e-05,
+      "loss": 1.061,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.710513749774151e-05,
+      "loss": 1.0049,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7102944446682393e-05,
+      "loss": 1.0306,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.710075070594782e-05,
+      "loss": 1.0071,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.709855627575079e-05,
+      "loss": 1.027,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7096361156304377e-05,
+      "loss": 1.0496,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7094165347821724e-05,
+      "loss": 1.0401,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7091968850516032e-05,
+      "loss": 0.9722,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7089771664600584e-05,
+      "loss": 1.061,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.708757379028871e-05,
+      "loss": 1.0027,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.708537522779382e-05,
+      "loss": 1.0335,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7083175977329394e-05,
+      "loss": 1.0795,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7080976039108964e-05,
+      "loss": 1.0751,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.707877541334615e-05,
+      "loss": 1.0595,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7076574100254614e-05,
+      "loss": 1.0326,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7074372100048103e-05,
+      "loss": 1.0521,
+      "step": 2831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.707216941294042e-05,
+      "loss": 0.933,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706996603914545e-05,
+      "loss": 0.9707,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706776197887712e-05,
+      "loss": 0.9857,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7065557232349456e-05,
+      "loss": 1.0144,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7063351799776514e-05,
+      "loss": 1.0432,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7061145681372444e-05,
+      "loss": 1.0427,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7058938877351456e-05,
+      "loss": 1.0504,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7056731387927822e-05,
+      "loss": 1.0091,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705452321331588e-05,
+      "loss": 0.9858,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705231435373004e-05,
+      "loss": 1.0048,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7050104809384774e-05,
+      "loss": 1.0197,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7047894580494627e-05,
+      "loss": 1.0867,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.70456836672742e-05,
+      "loss": 1.0224,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7043472069938173e-05,
+      "loss": 1.0447,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.704125978870128e-05,
+      "loss": 1.043,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.703904682377833e-05,
+      "loss": 0.9971,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7036833175384192e-05,
+      "loss": 1.0856,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.703461884373381e-05,
+      "loss": 1.0232,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7032403829042182e-05,
+      "loss": 1.0244,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.703018813152439e-05,
+      "loss": 1.0807,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7027971751395563e-05,
+      "loss": 1.0397,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.702575468887091e-05,
+      "loss": 1.0346,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7023536944165697e-05,
+      "loss": 1.0454,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7021318517495265e-05,
+      "loss": 1.0486,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7019099409075014e-05,
+      "loss": 0.9793,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7016879619120415e-05,
+      "loss": 1.058,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7014659147847005e-05,
+      "loss": 1.0197,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7012437995470377e-05,
+      "loss": 1.0471,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701021616220621e-05,
+      "loss": 1.09,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7007993648270227e-05,
+      "loss": 0.9583,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7005770453878234e-05,
+      "loss": 0.9573,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7003546579246097e-05,
+      "loss": 1.0023,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7001322024589742e-05,
+      "loss": 1.046,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6999096790125175e-05,
+      "loss": 1.0425,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6996870876068455e-05,
+      "loss": 0.9199,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6994644282635713e-05,
+      "loss": 0.9775,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6992417010043144e-05,
+      "loss": 1.0153,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.699018905850701e-05,
+      "loss": 1.0356,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6987960428243637e-05,
+      "loss": 1.0704,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698573111946942e-05,
+      "loss": 1.1072,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6983501132400825e-05,
+      "loss": 0.9896,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6981270467254366e-05,
+      "loss": 1.1046,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6979039124246643e-05,
+      "loss": 1.0018,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6976807103594307e-05,
+      "loss": 0.9448,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6974574405514083e-05,
+      "loss": 1.0463,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.697234103022276e-05,
+      "loss": 1.0709,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6970106977937192e-05,
+      "loss": 1.004,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6967872248874303e-05,
+      "loss": 1.0939,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696563684325107e-05,
+      "loss": 0.9723,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696340076128455e-05,
+      "loss": 1.0542,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6961164003191862e-05,
+      "loss": 1.095,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6958926569190185e-05,
+      "loss": 1.0504,
+      "step": 2883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6956688459496767e-05,
+      "loss": 0.9904,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.695444967432893e-05,
+      "loss": 1.0126,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.695221021390404e-05,
+      "loss": 0.8912,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.694997007843955e-05,
+      "loss": 0.9562,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6947729268152972e-05,
+      "loss": 1.0684,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.694548778326188e-05,
+      "loss": 1.0369,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6943245623983918e-05,
+      "loss": 0.9846,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6941002790536786e-05,
+      "loss": 0.9739,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6938759283138268e-05,
+      "loss": 0.9608,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6936515102006193e-05,
+      "loss": 0.9896,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693427024735847e-05,
+      "loss": 0.9711,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6932024719413063e-05,
+      "loss": 1.035,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.692977851838801e-05,
+      "loss": 1.0263,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6927531644501408e-05,
+      "loss": 1.0544,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6925284097971427e-05,
+      "loss": 1.0366,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.692303587901629e-05,
+      "loss": 1.1243,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6920786987854296e-05,
+      "loss": 0.9392,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691853742470381e-05,
+      "loss": 1.0188,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691628718978325e-05,
+      "loss": 0.9955,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6914036283311114e-05,
+      "loss": 1.0163,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691178470550596e-05,
+      "loss": 1.0009,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.69095324565864e-05,
+      "loss": 0.9999,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6907279536771127e-05,
+      "loss": 1.0121,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6905025946278896e-05,
+      "loss": 1.009,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6902771685328524e-05,
+      "loss": 1.1198,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6900516754138884e-05,
+      "loss": 1.0102,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6898261152928933e-05,
+      "loss": 1.0161,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6896004881917677e-05,
+      "loss": 1.0664,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6893747941324197e-05,
+      "loss": 0.9647,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6891490331367636e-05,
+      "loss": 1.0117,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6889232052267203e-05,
+      "loss": 0.9731,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6886973104242162e-05,
+      "loss": 1.0336,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.688471348751186e-05,
+      "loss": 0.9753,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6882453202295694e-05,
+      "loss": 1.0819,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.688019224881313e-05,
+      "loss": 1.0154,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687793062728371e-05,
+      "loss": 1.0383,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6875668337927014e-05,
+      "loss": 1.046,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6873405380962718e-05,
+      "loss": 1.0332,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6871141756610544e-05,
+      "loss": 1.0367,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686887746509028e-05,
+      "loss": 1.0476,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6866612506621788e-05,
+      "loss": 1.035,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686434688142499e-05,
+      "loss": 1.012,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6862080589719863e-05,
+      "loss": 1.0327,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6859813631726465e-05,
+      "loss": 1.0727,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6857546007664908e-05,
+      "loss": 1.0004,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685527771775537e-05,
+      "loss": 0.9986,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6853008762218103e-05,
+      "loss": 1.0523,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6850739141273408e-05,
+      "loss": 0.9717,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.684846885514166e-05,
+      "loss": 0.9503,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.68461979040433e-05,
+      "loss": 1.0132,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6843926288198828e-05,
+      "loss": 1.0266,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6841654007828816e-05,
+      "loss": 1.0451,
+      "step": 2935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.683938106315389e-05,
+      "loss": 1.004,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6837107454394752e-05,
+      "loss": 1.0449,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.683483318177216e-05,
+      "loss": 0.9843,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6832558245506937e-05,
+      "loss": 1.0121,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6830282645819974e-05,
+      "loss": 1.0263,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.682800638293223e-05,
+      "loss": 0.9537,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6825729457064718e-05,
+      "loss": 0.945,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6823451868438522e-05,
+      "loss": 0.919,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6821173617274793e-05,
+      "loss": 1.0453,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6818894703794735e-05,
+      "loss": 1.0773,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6816615128219635e-05,
+      "loss": 1.0017,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6814334890770824e-05,
+      "loss": 1.0485,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.681205399166971e-05,
+      "loss": 1.0117,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6809772431137757e-05,
+      "loss": 1.0184,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6807490209396506e-05,
+      "loss": 1.0444,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6805207326667552e-05,
+      "loss": 0.9749,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6802923783172553e-05,
+      "loss": 1.0643,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6800639579133234e-05,
+      "loss": 0.9485,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.679835471477139e-05,
+      "loss": 1.0979,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6796069190308867e-05,
+      "loss": 1.028,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6793783005967593e-05,
+      "loss": 1.0272,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.679149616196954e-05,
+      "loss": 1.0537,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678920865853676e-05,
+      "loss": 0.9942,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678692049589136e-05,
+      "loss": 0.9786,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678463167425552e-05,
+      "loss": 0.9856,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6782342193851467e-05,
+      "loss": 0.9928,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6780052054901512e-05,
+      "loss": 1.0279,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.677776125762802e-05,
+      "loss": 0.9625,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6775469802253416e-05,
+      "loss": 1.0068,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.67731776890002e-05,
+      "loss": 1.0646,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6770884918090923e-05,
+      "loss": 1.0288,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.676859148974821e-05,
+      "loss": 0.9998,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6766297404194745e-05,
+      "loss": 1.0299,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.676400266165328e-05,
+      "loss": 1.0489,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6761707262346624e-05,
+      "loss": 1.0226,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.675941120649766e-05,
+      "loss": 0.9711,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.675711449432932e-05,
+      "loss": 1.0461,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6754817126064616e-05,
+      "loss": 1.0582,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6752519101926606e-05,
+      "loss": 0.9346,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6750220422138435e-05,
+      "loss": 1.0304,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6747921086923284e-05,
+      "loss": 1.032,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6745621096504424e-05,
+      "loss": 1.0447,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674332045110517e-05,
+      "loss": 1.1082,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6741019150948908e-05,
+      "loss": 1.0436,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6738717196259092e-05,
+      "loss": 0.9756,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6736414587259234e-05,
+      "loss": 0.9573,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673411132417291e-05,
+      "loss": 1.0311,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6731807407223755e-05,
+      "loss": 0.9319,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.672950283663548e-05,
+      "loss": 0.9899,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6727197612631853e-05,
+      "loss": 1.0351,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6724891735436697e-05,
+      "loss": 1.0646,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6722585205273912e-05,
+      "loss": 1.1241,
+      "step": 2987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6720278022367453e-05,
+      "loss": 1.0656,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6717970186941343e-05,
+      "loss": 0.984,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6715661699219664e-05,
+      "loss": 0.9405,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6713352559426566e-05,
+      "loss": 1.0239,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6711042767786257e-05,
+      "loss": 1.032,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.670873232452301e-05,
+      "loss": 0.9965,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6706421229861168e-05,
+      "loss": 1.0417,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6704109484025126e-05,
+      "loss": 0.9904,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6701797087239354e-05,
+      "loss": 1.0031,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.669948403972837e-05,
+      "loss": 1.0443,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6697170341716772e-05,
+      "loss": 0.9686,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.669485599342921e-05,
+      "loss": 0.9296,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6692540995090403e-05,
+      "loss": 1.007,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.669022534692513e-05,
+      "loss": 1.0421,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.668790904915823e-05,
+      "loss": 1.0033,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6685592102014615e-05,
+      "loss": 0.992,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6683274505719248e-05,
+      "loss": 1.0005,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6680956260497164e-05,
+      "loss": 1.0544,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6678637366573455e-05,
+      "loss": 1.0651,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6676317824173284e-05,
+      "loss": 1.0195,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.667399763352187e-05,
+      "loss": 1.0342,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.66716767948445e-05,
+      "loss": 0.9791,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.666935530836651e-05,
+      "loss": 1.0616,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6667033174313318e-05,
+      "loss": 0.9495,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6664710392910396e-05,
+      "loss": 1.0522,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.666238696438328e-05,
+      "loss": 1.0112,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6660062888957564e-05,
+      "loss": 0.9996,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6657738166858913e-05,
+      "loss": 0.9261,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665541279831305e-05,
+      "loss": 0.9989,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665308678354576e-05,
+      "loss": 1.011,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6650760122782898e-05,
+      "loss": 0.9969,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6648432816250366e-05,
+      "loss": 0.9958,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6646104864174147e-05,
+      "loss": 0.9565,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6643776266780276e-05,
+      "loss": 1.023,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664144702429485e-05,
+      "loss": 1.0439,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663911713694404e-05,
+      "loss": 0.9664,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663678660495406e-05,
+      "loss": 1.0052,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663445542855121e-05,
+      "loss": 0.9942,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663212360796183e-05,
+      "loss": 1.0768,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662979114341234e-05,
+      "loss": 1.0302,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662745803512921e-05,
+      "loss": 0.9483,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6625124283338982e-05,
+      "loss": 1.017,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662278988826826e-05,
+      "loss": 1.0455,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6620454850143698e-05,
+      "loss": 0.9664,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6618119169192027e-05,
+      "loss": 1.066,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6615782845640037e-05,
+      "loss": 1.0432,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661344587971457e-05,
+      "loss": 0.9939,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6611108271642547e-05,
+      "loss": 0.9863,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6608770021650945e-05,
+      "loss": 1.0321,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6606431129966793e-05,
+      "loss": 1.106,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6604091596817193e-05,
+      "loss": 1.0583,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6601751422429307e-05,
+      "loss": 1.1855,
+      "step": 3039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6599410607030363e-05,
+      "loss": 1.0872,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6597069150847643e-05,
+      "loss": 1.0679,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6594727054108498e-05,
+      "loss": 1.0955,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6592384317040343e-05,
+      "loss": 1.0386,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659004093987064e-05,
+      "loss": 1.0239,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6587696922826932e-05,
+      "loss": 1.042,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6585352266136814e-05,
+      "loss": 0.9808,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.658300697002795e-05,
+      "loss": 1.0478,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6580661034728055e-05,
+      "loss": 1.0126,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6578314460464913e-05,
+      "loss": 1.059,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6575967247466376e-05,
+      "loss": 0.9457,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6573619395960346e-05,
+      "loss": 1.0121,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657127090617479e-05,
+      "loss": 1.0315,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656892177833775e-05,
+      "loss": 1.0162,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656657201267731e-05,
+      "loss": 1.0104,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656422160942163e-05,
+      "loss": 1.0634,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6561870568798927e-05,
+      "loss": 1.091,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6559518891037477e-05,
+      "loss": 1.0546,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.655716657636562e-05,
+      "loss": 0.9729,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6554813625011773e-05,
+      "loss": 1.0186,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6552460037204382e-05,
+      "loss": 1.059,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6550105813171985e-05,
+      "loss": 0.9593,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6547750953143168e-05,
+      "loss": 0.9755,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654539545734658e-05,
+      "loss": 1.0841,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654303932601093e-05,
+      "loss": 1.0384,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6540682559364996e-05,
+      "loss": 0.91,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6538325157637614e-05,
+      "loss": 0.9623,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.653596712105768e-05,
+      "loss": 0.991,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.653360844985415e-05,
+      "loss": 0.949,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6531249144256043e-05,
+      "loss": 0.8497,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.652888920449245e-05,
+      "loss": 1.0436,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6526528630792502e-05,
+      "loss": 1.0336,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6524167423385414e-05,
+      "loss": 1.0322,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6521805582500447e-05,
+      "loss": 1.0565,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.651944310836693e-05,
+      "loss": 0.9762,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.651708000121425e-05,
+      "loss": 1.001,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6514716261271866e-05,
+      "loss": 0.9518,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.651235188876928e-05,
+      "loss": 1.0156,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6509986883936073e-05,
+      "loss": 1.0267,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.650762124700188e-05,
+      "loss": 1.0146,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.650525497819639e-05,
+      "loss": 1.0331,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.650288807774937e-05,
+      "loss": 1.0257,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6500520545890634e-05,
+      "loss": 1.0132,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649815238285007e-05,
+      "loss": 1.0742,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6495783588857605e-05,
+      "loss": 1.0673,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6493414164143254e-05,
+      "loss": 1.0214,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649104410893708e-05,
+      "loss": 1.0707,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6488673423469205e-05,
+      "loss": 0.9991,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.648630210796982e-05,
+      "loss": 0.9938,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.648393016266917e-05,
+      "loss": 1.0128,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6481557587797562e-05,
+      "loss": 0.9875,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6479184383585368e-05,
+      "loss": 1.0659,
+      "step": 3091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6476810550263023e-05,
+      "loss": 1.0309,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6474436088061016e-05,
+      "loss": 1.0107,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6472060997209898e-05,
+      "loss": 0.9821,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646968527794029e-05,
+      "loss": 1.0599,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6467308930482863e-05,
+      "loss": 1.0116,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6464931955068354e-05,
+      "loss": 1.0211,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6462554351927558e-05,
+      "loss": 1.0281,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6460176121291337e-05,
+      "loss": 1.0618,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6457797263390613e-05,
+      "loss": 1.0577,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6455417778456362e-05,
+      "loss": 1.0175,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6453037666719624e-05,
+      "loss": 1.0081,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6450656928411507e-05,
+      "loss": 1.0431,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6448275563763162e-05,
+      "loss": 1.0532,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6445893573005827e-05,
+      "loss": 1.0712,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644351095637078e-05,
+      "loss": 0.9367,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6441127714089362e-05,
+      "loss": 0.9984,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6438743846392987e-05,
+      "loss": 1.0535,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6436359353513112e-05,
+      "loss": 1.0271,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6433974235681274e-05,
+      "loss": 0.9849,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6431588493129057e-05,
+      "loss": 0.9409,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6429202126088112e-05,
+      "loss": 1.0085,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6426815134790143e-05,
+      "loss": 1.0415,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6424427519466925e-05,
+      "loss": 1.0754,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6422039280350287e-05,
+      "loss": 0.9301,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.641965041767212e-05,
+      "loss": 1.0265,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6417260931664373e-05,
+      "loss": 0.9928,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6414870822559064e-05,
+      "loss": 0.8977,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6412480090588264e-05,
+      "loss": 1.0484,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6410088735984103e-05,
+      "loss": 1.0277,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6407696758978774e-05,
+      "loss": 1.0504,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6405304159804534e-05,
+      "loss": 1.0215,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6402910938693702e-05,
+      "loss": 1.0317,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6400517095878644e-05,
+      "loss": 1.0304,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.63981226315918e-05,
+      "loss": 1.0742,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6395727546065665e-05,
+      "loss": 1.0922,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6393331839532795e-05,
+      "loss": 1.0693,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6390935512225806e-05,
+      "loss": 0.991,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6388538564377373e-05,
+      "loss": 1.0444,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6386140996220232e-05,
+      "loss": 0.9871,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6383742807987187e-05,
+      "loss": 1.025,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6381343999911088e-05,
+      "loss": 0.9825,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6378944572224855e-05,
+      "loss": 0.9264,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6376544525161463e-05,
+      "loss": 1.0622,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6374143858953954e-05,
+      "loss": 1.0534,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6371742573835426e-05,
+      "loss": 1.004,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6369340670039032e-05,
+      "loss": 1.0472,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.636693814779799e-05,
+      "loss": 0.9851,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6364535007345583e-05,
+      "loss": 1.0063,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6362131248915145e-05,
+      "loss": 1.019,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.635972687274008e-05,
+      "loss": 0.9977,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6357321879053833e-05,
+      "loss": 1.0137,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6354916268089938e-05,
+      "loss": 0.9946,
+      "step": 3143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6352510040081962e-05,
+      "loss": 1.0485,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6350103195263547e-05,
+      "loss": 1.1139,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634769573386839e-05,
+      "loss": 0.998,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6345287656130246e-05,
+      "loss": 1.0455,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634287896228294e-05,
+      "loss": 1.0508,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634046965256034e-05,
+      "loss": 1.0182,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6338059727196386e-05,
+      "loss": 0.9547,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6335649186425078e-05,
+      "loss": 0.9131,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6333238030480473e-05,
+      "loss": 0.9223,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6330826259596683e-05,
+      "loss": 1.0275,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6328413874007884e-05,
+      "loss": 0.9447,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6326000873948313e-05,
+      "loss": 1.0325,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6323587259652267e-05,
+      "loss": 1.051,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.63211730313541e-05,
+      "loss": 1.0368,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6318758189288227e-05,
+      "loss": 0.9977,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.631634273368912e-05,
+      "loss": 1.0801,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6313926664791316e-05,
+      "loss": 1.0814,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6311509982829407e-05,
+      "loss": 1.0103,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6309092688038047e-05,
+      "loss": 1.0318,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6306674780651944e-05,
+      "loss": 0.9879,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6304256260905872e-05,
+      "loss": 0.9851,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6301837129034666e-05,
+      "loss": 1.0737,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6299417385273216e-05,
+      "loss": 1.0493,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6296997029856466e-05,
+      "loss": 0.9647,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629457606301943e-05,
+      "loss": 0.9647,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6292154484997178e-05,
+      "loss": 1.0651,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6289732296024837e-05,
+      "loss": 1.0247,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6287309496337594e-05,
+      "loss": 0.9902,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6284886086170697e-05,
+      "loss": 0.9745,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628246206575945e-05,
+      "loss": 0.9124,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628003743533922e-05,
+      "loss": 0.9962,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627761219514543e-05,
+      "loss": 1.0038,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6275186345413566e-05,
+      "loss": 1.0918,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627275988637917e-05,
+      "loss": 1.0002,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627033281827785e-05,
+      "loss": 0.9543,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6267905141345256e-05,
+      "loss": 1.0873,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6265476855817116e-05,
+      "loss": 0.974,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6263047961929208e-05,
+      "loss": 0.9721,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6260618459917366e-05,
+      "loss": 0.8997,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6258188350017497e-05,
+      "loss": 1.0233,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6255757632465553e-05,
+      "loss": 1.08,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625332630749755e-05,
+      "loss": 1.0476,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625089437534956e-05,
+      "loss": 1.0129,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6248461836257718e-05,
+      "loss": 1.0151,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624602869045822e-05,
+      "loss": 0.9489,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624359493818731e-05,
+      "loss": 0.9565,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624116057968131e-05,
+      "loss": 0.9896,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6238725615176576e-05,
+      "loss": 1.0049,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6236290044909543e-05,
+      "loss": 0.9913,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6233853869116697e-05,
+      "loss": 0.9341,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6231417088034585e-05,
+      "loss": 0.9729,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6228979701899814e-05,
+      "loss": 1.0577,
+      "step": 3195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.622654171094904e-05,
+      "loss": 1.0709,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6224103115418987e-05,
+      "loss": 1.0196,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6221663915546437e-05,
+      "loss": 0.919,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.621922411156823e-05,
+      "loss": 1.0293,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6216783703721265e-05,
+      "loss": 1.0227,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6214342692242494e-05,
+      "loss": 0.947,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6211901077368937e-05,
+      "loss": 0.9728,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6209458859337662e-05,
+      "loss": 1.0105,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.620701603838581e-05,
+      "loss": 0.925,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6204572614750564e-05,
+      "loss": 1.0627,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6202128588669177e-05,
+      "loss": 1.0324,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.619968396037896e-05,
+      "loss": 0.9412,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.619723873011727e-05,
+      "loss": 1.0143,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6194792898121537e-05,
+      "loss": 0.9504,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6192346464629247e-05,
+      "loss": 0.9406,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618989942987794e-05,
+      "loss": 0.927,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6187451794105212e-05,
+      "loss": 1.0306,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6185003557548728e-05,
+      "loss": 0.9424,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61825547204462e-05,
+      "loss": 0.9896,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6180105283035407e-05,
+      "loss": 0.9768,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6177655245554177e-05,
+      "loss": 0.9696,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6175204608240405e-05,
+      "loss": 0.9778,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.617275337133204e-05,
+      "loss": 1.0682,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6170301535067094e-05,
+      "loss": 0.9579,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6167849099683623e-05,
+      "loss": 1.0337,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.616539606541976e-05,
+      "loss": 1.0342,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6162942432513687e-05,
+      "loss": 0.9946,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6160488201203643e-05,
+      "loss": 0.9532,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6158033371727924e-05,
+      "loss": 0.9847,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6155577944324895e-05,
+      "loss": 1.0668,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6153121919232962e-05,
+      "loss": 1.0254,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6150665296690605e-05,
+      "loss": 1.0522,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.614820807693635e-05,
+      "loss": 1.0016,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6145750260208786e-05,
+      "loss": 1.0961,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6143291846746563e-05,
+      "loss": 0.98,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6140832836788386e-05,
+      "loss": 0.9444,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.613837323057301e-05,
+      "loss": 0.947,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.613591302833927e-05,
+      "loss": 0.9753,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6133452230326035e-05,
+      "loss": 1.0119,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.613099083677224e-05,
+      "loss": 1.0117,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6128528847916883e-05,
+      "loss": 0.9642,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6126066263999012e-05,
+      "loss": 1.0282,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6123603085257746e-05,
+      "loss": 1.0266,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6121139311932243e-05,
+      "loss": 0.9593,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6118674944261732e-05,
+      "loss": 0.9825,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61162099824855e-05,
+      "loss": 1.0228,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6113744426842882e-05,
+      "loss": 1.0221,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6111278277573275e-05,
+      "loss": 0.9735,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6108811534916137e-05,
+      "loss": 1.04,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6106344199110987e-05,
+      "loss": 1.0272,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6103876270397387e-05,
+      "loss": 0.9664,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6101407749014973e-05,
+      "loss": 0.9475,
+      "step": 3247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.609893863520343e-05,
+      "loss": 1.1213,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60964689292025e-05,
+      "loss": 0.9417,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.609399863125198e-05,
+      "loss": 1.0858,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6091527741591742e-05,
+      "loss": 0.9828,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6089056260461687e-05,
+      "loss": 0.9882,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60865841881018e-05,
+      "loss": 1.0371,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6084111524752107e-05,
+      "loss": 0.9738,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60816382706527e-05,
+      "loss": 1.0462,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607916442604372e-05,
+      "loss": 0.9311,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607668999116537e-05,
+      "loss": 1.0099,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6074214966257914e-05,
+      "loss": 0.9393,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607173935156167e-05,
+      "loss": 1.0087,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6069263147317015e-05,
+      "loss": 1.0309,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6066786353764373e-05,
+      "loss": 0.979,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6064308971144236e-05,
+      "loss": 1.0244,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.606183099969716e-05,
+      "loss": 0.9464,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.605935243966374e-05,
+      "loss": 0.9905,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.605687329128464e-05,
+      "loss": 0.9584,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6054393554800574e-05,
+      "loss": 1.0131,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.605191323045233e-05,
+      "loss": 1.1139,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.604943231848072e-05,
+      "loss": 1.0147,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.604695081912665e-05,
+      "loss": 1.0224,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.604446873263106e-05,
+      "loss": 1.0291,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6041986059234954e-05,
+      "loss": 0.9208,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6039502799179394e-05,
+      "loss": 0.9646,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6037018952705493e-05,
+      "loss": 0.9979,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6034534520054435e-05,
+      "loss": 1.007,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.603204950146744e-05,
+      "loss": 1.0492,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60295638971858e-05,
+      "loss": 0.972,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6027077707450867e-05,
+      "loss": 1.021,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.602459093250403e-05,
+      "loss": 1.0369,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6022103572586756e-05,
+      "loss": 1.038,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.601961562794056e-05,
+      "loss": 0.9241,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6017127098807017e-05,
+      "loss": 1.0406,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.601463798542775e-05,
+      "loss": 1.0468,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6012148288044443e-05,
+      "loss": 0.9469,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6009658006898848e-05,
+      "loss": 0.996,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6007167142232757e-05,
+      "loss": 1.0286,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.600467569428803e-05,
+      "loss": 0.9595,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.600218366330657e-05,
+      "loss": 1.0299,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.599969104953036e-05,
+      "loss": 1.0142,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5997197853201412e-05,
+      "loss": 0.9819,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.599470407456182e-05,
+      "loss": 0.9962,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5992209713853717e-05,
+      "loss": 0.9854,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5989714771319297e-05,
+      "loss": 0.9983,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5987219247200813e-05,
+      "loss": 0.9604,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5984723141740578e-05,
+      "loss": 1.0382,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5982226455180946e-05,
+      "loss": 0.9267,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.597972918776435e-05,
+      "loss": 0.9919,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5977231339733258e-05,
+      "loss": 1.0222,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5974732911330208e-05,
+      "loss": 1.0259,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5972233902797788e-05,
+      "loss": 1.069,
+      "step": 3299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5969734314378654e-05,
+      "loss": 1.0522,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5967234146315497e-05,
+      "loss": 0.9908,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5964733398851078e-05,
+      "loss": 0.9808,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5962232072228216e-05,
+      "loss": 0.9709,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5959730166689783e-05,
+      "loss": 1.057,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5957227682478705e-05,
+      "loss": 0.9137,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5954724619837966e-05,
+      "loss": 1.0613,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5952220979010605e-05,
+      "loss": 0.9677,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5949716760239722e-05,
+      "loss": 1.0385,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.594721196376847e-05,
+      "loss": 1.0141,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5944706589840046e-05,
+      "loss": 0.9391,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5942200638697728e-05,
+      "loss": 1.0222,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5939694110584833e-05,
+      "loss": 1.0731,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5937187005744735e-05,
+      "loss": 0.9747,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.593467932442087e-05,
+      "loss": 0.9882,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5932171066856724e-05,
+      "loss": 0.9523,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5929662233295846e-05,
+      "loss": 0.9824,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5927152823981827e-05,
+      "loss": 0.9357,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5924642839158334e-05,
+      "loss": 1.004,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5922132279069074e-05,
+      "loss": 1.0267,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591962114395781e-05,
+      "loss": 1.0073,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591710943406838e-05,
+      "loss": 0.9453,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5914597149644654e-05,
+      "loss": 1.0131,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5912084290930568e-05,
+      "loss": 0.9686,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5909570858170115e-05,
+      "loss": 0.9385,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5907056851607337e-05,
+      "loss": 0.9919,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5904542271486346e-05,
+      "loss": 1.0026,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.59020271180513e-05,
+      "loss": 0.9757,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5899511391546403e-05,
+      "loss": 1.0158,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.589699509221593e-05,
+      "loss": 1.0197,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5894478220304215e-05,
+      "loss": 1.0372,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5891960776055625e-05,
+      "loss": 1.0848,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5889442759714603e-05,
+      "loss": 0.9864,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5886924171525643e-05,
+      "loss": 1.021,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5884405011733294e-05,
+      "loss": 1.0052,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.588188528058215e-05,
+      "loss": 0.9698,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587936497831688e-05,
+      "loss": 1.0023,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5876844105182194e-05,
+      "loss": 0.9104,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5874322661422856e-05,
+      "loss": 1.0051,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.58718006472837e-05,
+      "loss": 1.0054,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5869278063009602e-05,
+      "loss": 1.0315,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.58667549088455e-05,
+      "loss": 1.1085,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.586423118503638e-05,
+      "loss": 0.9587,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5861706891827293e-05,
+      "loss": 1.0704,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585918202946334e-05,
+      "loss": 0.9854,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5856656598189673e-05,
+      "loss": 1.0519,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5854130598251514e-05,
+      "loss": 1.0201,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5851604029894124e-05,
+      "loss": 0.9911,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5849076893362822e-05,
+      "loss": 0.9306,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.584654918890299e-05,
+      "loss": 1.0178,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.584402091676006e-05,
+      "loss": 1.0651,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5841492077179523e-05,
+      "loss": 1.0485,
+      "step": 3351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5838962670406918e-05,
+      "loss": 0.9423,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5836432696687846e-05,
+      "loss": 0.9592,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5833902156267956e-05,
+      "loss": 0.948,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5831371049392957e-05,
+      "loss": 1.1011,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.582883937630862e-05,
+      "loss": 0.9766,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5826307137260754e-05,
+      "loss": 0.9805,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5823774332495236e-05,
+      "loss": 1.0433,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5821240962257996e-05,
+      "loss": 1.0451,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.581870702679501e-05,
+      "loss": 0.9881,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.581617252635232e-05,
+      "loss": 0.959,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.581363746117602e-05,
+      "loss": 0.9477,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5811101831512262e-05,
+      "loss": 1.0723,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580856563760724e-05,
+      "loss": 0.9648,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580602887970721e-05,
+      "loss": 1.0004,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5803491558058486e-05,
+      "loss": 0.9837,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580095367290744e-05,
+      "loss": 0.9239,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.579841522450049e-05,
+      "loss": 1.0474,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5795876213084114e-05,
+      "loss": 0.9538,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5793336638904838e-05,
+      "loss": 1.0491,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5790796502209253e-05,
+      "loss": 1.0073,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.578825580324399e-05,
+      "loss": 0.9914,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5785714542255753e-05,
+      "loss": 1.012,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5783172719491288e-05,
+      "loss": 0.989,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5780630335197397e-05,
+      "loss": 1.0014,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577808738962094e-05,
+      "loss": 1.0426,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5775543883008828e-05,
+      "loss": 1.0403,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577299981560803e-05,
+      "loss": 1.0118,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5770455187665565e-05,
+      "loss": 0.9742,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5767909999428513e-05,
+      "loss": 0.9675,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5765364251144003e-05,
+      "loss": 0.9395,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.576281794305922e-05,
+      "loss": 1.0043,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5760271075421402e-05,
+      "loss": 1.1013,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.575772364847784e-05,
+      "loss": 1.002,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5755175662475886e-05,
+      "loss": 1.0049,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.575262711766294e-05,
+      "loss": 1.0488,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.575007801428646e-05,
+      "loss": 0.9203,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5747528352593956e-05,
+      "loss": 0.9475,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5744978132832993e-05,
+      "loss": 1.0493,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.574242735525119e-05,
+      "loss": 1.0229,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5739876020096215e-05,
+      "loss": 1.0304,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5737324127615808e-05,
+      "loss": 0.9999,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.573477167805774e-05,
+      "loss": 1.0076,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5732218671669847e-05,
+      "loss": 1.0565,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.572966510870002e-05,
+      "loss": 1.062,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5727110989396205e-05,
+      "loss": 1.0597,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5724556314006398e-05,
+      "loss": 0.9192,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5722001082778645e-05,
+      "loss": 0.9737,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5719445295961064e-05,
+      "loss": 1.027,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5716888953801805e-05,
+      "loss": 0.9422,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5714332056549086e-05,
+      "loss": 0.9526,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5711774604451168e-05,
+      "loss": 1.0198,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.570921659775638e-05,
+      "loss": 1.0688,
+      "step": 3403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5706658036713093e-05,
+      "loss": 1.0399,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5704098921569738e-05,
+      "loss": 1.0001,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5701539252574795e-05,
+      "loss": 0.9497,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.56989790299768e-05,
+      "loss": 1.0295,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5696418254024344e-05,
+      "loss": 1.004,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.569385692496608e-05,
+      "loss": 1.0372,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.569129504305069e-05,
+      "loss": 1.014,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5688732608526937e-05,
+      "loss": 0.9912,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.568616962164362e-05,
+      "loss": 1.0155,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5683606082649602e-05,
+      "loss": 0.994,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5681041991793788e-05,
+      "loss": 1.0466,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5678477349325155e-05,
+      "loss": 1.0348,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.567591215549271e-05,
+      "loss": 0.9965,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5673346410545537e-05,
+      "loss": 0.9598,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.567078011473276e-05,
+      "loss": 0.9193,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.566821326830355e-05,
+      "loss": 1.0674,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5665645871507152e-05,
+      "loss": 1.0058,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5663077924592847e-05,
+      "loss": 1.0312,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5660509427809973e-05,
+      "loss": 1.0092,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.565794038140793e-05,
+      "loss": 0.9881,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.565537078563616e-05,
+      "loss": 1.0966,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5652800640744165e-05,
+      "loss": 1.0685,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.56502299469815e-05,
+      "loss": 1.0058,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5647658704597772e-05,
+      "loss": 1.0717,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.564508691384264e-05,
+      "loss": 0.9405,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5642514574965812e-05,
+      "loss": 0.9909,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5639941688217063e-05,
+      "loss": 0.881,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563736825384621e-05,
+      "loss": 0.9539,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5634794272103126e-05,
+      "loss": 0.9482,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563221974323774e-05,
+      "loss": 0.9886,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562964466750003e-05,
+      "loss": 0.9709,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5627069045140022e-05,
+      "loss": 0.9464,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562449287640781e-05,
+      "loss": 0.9706,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5621916161553524e-05,
+      "loss": 0.9187,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5619338900827368e-05,
+      "loss": 0.9145,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5616761094479572e-05,
+      "loss": 0.9059,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5614182742760448e-05,
+      "loss": 0.9549,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5611603845920335e-05,
+      "loss": 1.0212,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5609024404209643e-05,
+      "loss": 1.0193,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5606444417878826e-05,
+      "loss": 1.0791,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5603863887178393e-05,
+      "loss": 1.0298,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5601282812358908e-05,
+      "loss": 0.9884,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5598701193670983e-05,
+      "loss": 0.9503,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5596119031365288e-05,
+      "loss": 0.9953,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559353632569254e-05,
+      "loss": 1.0001,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5590953076903522e-05,
+      "loss": 1.0291,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5588369285249048e-05,
+      "loss": 0.9522,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5585784950980002e-05,
+      "loss": 1.0212,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5583200074347318e-05,
+      "loss": 0.9755,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5580614655601976e-05,
+      "loss": 0.9966,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.557802869499501e-05,
+      "loss": 0.9798,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.557544219277752e-05,
+      "loss": 0.9966,
+      "step": 3455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5572855149200637e-05,
+      "loss": 1.011,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5570267564515566e-05,
+      "loss": 1.023,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5567679438973543e-05,
+      "loss": 1.0089,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5565090772825872e-05,
+      "loss": 1.0028,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5562501566323906e-05,
+      "loss": 0.9884,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.555991181971905e-05,
+      "loss": 1.0502,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.555732153326276e-05,
+      "loss": 1.0227,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5554730707206545e-05,
+      "loss": 1.0371,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5552139341801965e-05,
+      "loss": 1.0284,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554954743730064e-05,
+      "loss": 1.0101,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554695499395423e-05,
+      "loss": 0.9505,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5544362012014457e-05,
+      "loss": 1.0002,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5541768491733092e-05,
+      "loss": 1.041,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5539174433361955e-05,
+      "loss": 1.0645,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5536579837152927e-05,
+      "loss": 0.8979,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5533984703357935e-05,
+      "loss": 0.9655,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5531389032228955e-05,
+      "loss": 0.985,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552879282401802e-05,
+      "loss": 1.0505,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552619607897722e-05,
+      "loss": 1.0327,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5523598797358683e-05,
+      "loss": 0.9922,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.55210009794146e-05,
+      "loss": 1.0844,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.551840262539722e-05,
+      "loss": 1.0736,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5515803735558827e-05,
+      "loss": 0.9662,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5513204310151766e-05,
+      "loss": 0.9983,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5510604349428438e-05,
+      "loss": 1.0024,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5508003853641288e-05,
+      "loss": 1.0087,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.550540282304282e-05,
+      "loss": 1.0176,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5502801257885584e-05,
+      "loss": 0.9809,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.550019915842218e-05,
+      "loss": 0.9863,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5497596524905275e-05,
+      "loss": 0.9885,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.549499335758757e-05,
+      "loss": 1.0814,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5492389656721826e-05,
+      "loss": 0.933,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.548978542256086e-05,
+      "loss": 1.1029,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5487180655357525e-05,
+      "loss": 0.9662,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5484575355364744e-05,
+      "loss": 0.9859,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5481969522835485e-05,
+      "loss": 0.9791,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5479363158022763e-05,
+      "loss": 0.9768,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5476756261179652e-05,
+      "loss": 0.9964,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.547414883255927e-05,
+      "loss": 0.9139,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5471540872414793e-05,
+      "loss": 1.0135,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546893238099945e-05,
+      "loss": 0.9756,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5466323358566513e-05,
+      "loss": 0.9858,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5463713805369312e-05,
+      "loss": 1.0402,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5461103721661227e-05,
+      "loss": 1.0514,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5458493107695688e-05,
+      "loss": 0.9734,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.545588196372618e-05,
+      "loss": 0.9885,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5453270290006237e-05,
+      "loss": 0.9967,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5450658086789448e-05,
+      "loss": 1.0016,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.544804535432945e-05,
+      "loss": 0.9932,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5445432092879924e-05,
+      "loss": 0.9989,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.544281830269462e-05,
+      "loss": 0.9581,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5440203984027323e-05,
+      "loss": 1.0339,
+      "step": 3507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5437589137131882e-05,
+      "loss": 0.9341,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.543497376226218e-05,
+      "loss": 1.0598,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5432357859672177e-05,
+      "loss": 0.9768,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.542974142961586e-05,
+      "loss": 0.9246,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.542712447234728e-05,
+      "loss": 1.0936,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5424506988120533e-05,
+      "loss": 0.9935,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.542188897718977e-05,
+      "loss": 1.0156,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5419270439809198e-05,
+      "loss": 1.0275,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5416651376233062e-05,
+      "loss": 1.0659,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.541403178671567e-05,
+      "loss": 1.0523,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5411411671511376e-05,
+      "loss": 1.1046,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.540879103087458e-05,
+      "loss": 0.9943,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5406169865059747e-05,
+      "loss": 1.0382,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5403548174321383e-05,
+      "loss": 0.9187,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5400925958914045e-05,
+      "loss": 1.0261,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.539830321909234e-05,
+      "loss": 0.9956,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5395679955110927e-05,
+      "loss": 1.0601,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.539305616722453e-05,
+      "loss": 0.963,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.53904318556879e-05,
+      "loss": 0.9479,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5387807020755846e-05,
+      "loss": 0.932,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5385181662683244e-05,
+      "loss": 0.9773,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5382555781725004e-05,
+      "loss": 1.0146,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5379929378136088e-05,
+      "loss": 1.0558,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5377302452171515e-05,
+      "loss": 1.0249,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5374675004086353e-05,
+      "loss": 1.004,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.537204703413572e-05,
+      "loss": 0.927,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5369418542574782e-05,
+      "loss": 1.1215,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5366789529658755e-05,
+      "loss": 1.0356,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.536415999564292e-05,
+      "loss": 1.0046,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.536152994078258e-05,
+      "loss": 1.1005,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5358899365333123e-05,
+      "loss": 0.9492,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.535626826954996e-05,
+      "loss": 1.0378,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5353636653688563e-05,
+      "loss": 0.9787,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5351004518004463e-05,
+      "loss": 0.9938,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534837186275322e-05,
+      "loss": 1.0275,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5345738688190466e-05,
+      "loss": 1.1008,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5343104994571877e-05,
+      "loss": 0.993,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534047078215317e-05,
+      "loss": 1.0042,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.533783605119012e-05,
+      "loss": 1.0086,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5335200801938555e-05,
+      "loss": 0.9594,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5332565034654344e-05,
+      "loss": 1.0193,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5329928749593425e-05,
+      "loss": 0.9946,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5327291947011763e-05,
+      "loss": 0.9664,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5324654627165386e-05,
+      "loss": 0.9233,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5322016790310373e-05,
+      "loss": 1.0189,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5319378436702846e-05,
+      "loss": 1.0701,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5316739566598985e-05,
+      "loss": 1.0114,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5314100180255018e-05,
+      "loss": 1.0121,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.531146027792722e-05,
+      "loss": 0.9768,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5308819859871914e-05,
+      "loss": 1.0471,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.530617892634548e-05,
+      "loss": 1.0364,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.530353747760435e-05,
+      "loss": 1.0251,
+      "step": 3559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5300895513904993e-05,
+      "loss": 0.9053,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5298253035503942e-05,
+      "loss": 1.0768,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529561004265777e-05,
+      "loss": 1.0731,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5292966535623105e-05,
+      "loss": 0.9633,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5290322514656624e-05,
+      "loss": 0.9562,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5287677980015056e-05,
+      "loss": 0.9662,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5285032931955177e-05,
+      "loss": 1.0148,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5282387370733814e-05,
+      "loss": 0.9291,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527974129660784e-05,
+      "loss": 0.9868,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5277094709834183e-05,
+      "loss": 1.0397,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527444761066982e-05,
+      "loss": 0.9857,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5271799999371774e-05,
+      "loss": 0.9918,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5269151876197127e-05,
+      "loss": 1.0103,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5266503241402996e-05,
+      "loss": 1.0488,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5263854095246557e-05,
+      "loss": 1.0375,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5261204437985043e-05,
+      "loss": 0.9914,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5258554269875716e-05,
+      "loss": 0.9894,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5255903591175907e-05,
+      "loss": 0.9811,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5253252402142989e-05,
+      "loss": 0.9639,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5250600703034384e-05,
+      "loss": 1.0485,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5247948494107566e-05,
+      "loss": 0.9511,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5245295775620053e-05,
+      "loss": 1.0162,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5242642547829416e-05,
+      "loss": 0.9362,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5239988810993278e-05,
+      "loss": 0.9965,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.523733456536931e-05,
+      "loss": 0.9777,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5234679811215233e-05,
+      "loss": 0.9966,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5232024548788813e-05,
+      "loss": 0.9813,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5229368778347868e-05,
+      "loss": 1.0311,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5226712500150267e-05,
+      "loss": 1.0268,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5224055714453926e-05,
+      "loss": 1.0057,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5221398421516816e-05,
+      "loss": 0.9854,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5218740621596944e-05,
+      "loss": 1.0132,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5216082314952383e-05,
+      "loss": 1.0874,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5213423501841238e-05,
+      "loss": 1.0114,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.521076418252168e-05,
+      "loss": 0.9438,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5208104357251921e-05,
+      "loss": 1.0193,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5205444026290218e-05,
+      "loss": 0.9811,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5202783189894881e-05,
+      "loss": 1.0145,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5200121848324276e-05,
+      "loss": 0.9553,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5197460001836807e-05,
+      "loss": 0.9715,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5194797650690926e-05,
+      "loss": 1.0081,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5192134795145153e-05,
+      "loss": 0.9268,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5189471435458032e-05,
+      "loss": 1.0089,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5186807571888173e-05,
+      "loss": 1.0067,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5184143204694231e-05,
+      "loss": 1.0007,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5181478334134904e-05,
+      "loss": 0.966,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5178812960468945e-05,
+      "loss": 1.0503,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5176147083955153e-05,
+      "loss": 0.9314,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5173480704852379e-05,
+      "loss": 1.0065,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5170813823419523e-05,
+      "loss": 1.0342,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5168146439915525e-05,
+      "loss": 0.9728,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5165478554599385e-05,
+      "loss": 1.028,
+      "step": 3611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5162810167730144e-05,
+      "loss": 1.0226,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5160141279566895e-05,
+      "loss": 1.089,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5157471890368785e-05,
+      "loss": 0.9254,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5154802000394998e-05,
+      "loss": 1.0179,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5152131609904773e-05,
+      "loss": 1.047,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5149460719157398e-05,
+      "loss": 1.0677,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5146789328412213e-05,
+      "loss": 0.9471,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5144117437928595e-05,
+      "loss": 1.0704,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5141445047965984e-05,
+      "loss": 1.0284,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5138772158783854e-05,
+      "loss": 1.0654,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5136098770641741e-05,
+      "loss": 0.9425,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5133424883799221e-05,
+      "loss": 0.9789,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.513075049851592e-05,
+      "loss": 0.9627,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5128075615051516e-05,
+      "loss": 1.0177,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5125400233665728e-05,
+      "loss": 0.9592,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5122724354618332e-05,
+      "loss": 1.0717,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5120047978169146e-05,
+      "loss": 1.0219,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5117371104578036e-05,
+      "loss": 1.0779,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5114693734104926e-05,
+      "loss": 1.0739,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5112015867009775e-05,
+      "loss": 1.0314,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5109337503552594e-05,
+      "loss": 0.9043,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5106658643993455e-05,
+      "loss": 1.017,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5103979288592454e-05,
+      "loss": 0.9736,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5101299437609758e-05,
+      "loss": 0.9802,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5098619091305571e-05,
+      "loss": 1.0183,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5095938249940145e-05,
+      "loss": 0.9512,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5093256913773786e-05,
+      "loss": 0.9074,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5090575083066838e-05,
+      "loss": 1.0144,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.50878927580797e-05,
+      "loss": 1.038,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5085209939072822e-05,
+      "loss": 1.0815,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5082526626306698e-05,
+      "loss": 1.0072,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5079842820041867e-05,
+      "loss": 1.0547,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5077158520538921e-05,
+      "loss": 0.9922,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5074473728058493e-05,
+      "loss": 1.0164,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5071788442861277e-05,
+      "loss": 0.9949,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5069102665207998e-05,
+      "loss": 0.9471,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5066416395359444e-05,
+      "loss": 0.9724,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.506372963357644e-05,
+      "loss": 1.0034,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5061042380119864e-05,
+      "loss": 1.0109,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.505835463525064e-05,
+      "loss": 1.0348,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5055666399229743e-05,
+      "loss": 0.9973,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5052977672318187e-05,
+      "loss": 0.8813,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5050288454777047e-05,
+      "loss": 1.0642,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5047598746867433e-05,
+      "loss": 0.9621,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.504490854885051e-05,
+      "loss": 1.0181,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.504221786098749e-05,
+      "loss": 1.0248,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5039526683539627e-05,
+      "loss": 1.0226,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5036835016768229e-05,
+      "loss": 0.9898,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5034142860934649e-05,
+      "loss": 0.962,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5031450216300285e-05,
+      "loss": 0.9198,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5028757083126594e-05,
+      "loss": 0.9924,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.502606346167506e-05,
+      "loss": 0.9305,
+      "step": 3663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5023369352207229e-05,
+      "loss": 1.0619,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5020674754984697e-05,
+      "loss": 0.9685,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5017979670269096e-05,
+      "loss": 0.991,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5015284098322114e-05,
+      "loss": 0.987,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.501258803940548e-05,
+      "loss": 1.0369,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5009891493780978e-05,
+      "loss": 0.9914,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.500719446171043e-05,
+      "loss": 0.9476,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5004496943455716e-05,
+      "loss": 0.9329,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.500179893927875e-05,
+      "loss": 0.9532,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4999100449441506e-05,
+      "loss": 0.9459,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4996401474205997e-05,
+      "loss": 0.943,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4993702013834286e-05,
+      "loss": 0.9308,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4991002068588484e-05,
+      "loss": 1.0198,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4988301638730747e-05,
+      "loss": 1.0123,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4985600724523282e-05,
+      "loss": 0.9774,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4982899326228335e-05,
+      "loss": 1.0067,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4980197444108205e-05,
+      "loss": 0.9835,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4977495078425239e-05,
+      "loss": 1.0438,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4974792229441826e-05,
+      "loss": 1.0664,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4972088897420412e-05,
+      "loss": 1.003,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4969385082623473e-05,
+      "loss": 1.0284,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4966680785313548e-05,
+      "loss": 1.0445,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4963976005753216e-05,
+      "loss": 0.9567,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.49612707442051e-05,
+      "loss": 0.954,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4958565000931877e-05,
+      "loss": 0.8835,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4955858776196262e-05,
+      "loss": 0.953,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4953152070261027e-05,
+      "loss": 1.007,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4950444883388984e-05,
+      "loss": 0.9095,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.494773721584299e-05,
+      "loss": 1.0032,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4945029067885956e-05,
+      "loss": 0.9487,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4942320439780833e-05,
+      "loss": 0.9809,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4939611331790617e-05,
+      "loss": 0.9877,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4936901744178367e-05,
+      "loss": 0.9913,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4934191677207162e-05,
+      "loss": 1.0195,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4931481131140149e-05,
+      "loss": 0.9745,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4928770106240511e-05,
+      "loss": 1.0381,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4926058602771484e-05,
+      "loss": 1.0185,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4923346620996346e-05,
+      "loss": 1.0049,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4920634161178424e-05,
+      "loss": 0.9963,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.491792122358109e-05,
+      "loss": 0.9985,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4915207808467756e-05,
+      "loss": 0.9848,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4912493916101893e-05,
+      "loss": 1.0791,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4909779546747011e-05,
+      "loss": 0.9916,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4907064700666669e-05,
+      "loss": 1.0022,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4904349378124467e-05,
+      "loss": 1.0392,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4901633579384058e-05,
+      "loss": 1.084,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.489891730470914e-05,
+      "loss": 0.984,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4896200554363446e-05,
+      "loss": 0.8527,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4893483328610778e-05,
+      "loss": 0.8703,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.489076562771496e-05,
+      "loss": 1.008,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.488804745193988e-05,
+      "loss": 1.0122,
+      "step": 3714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4885328801549461e-05,
+      "loss": 1.0245,
+      "step": 3715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4882609676807675e-05,
+      "loss": 1.0718,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4879890077978547e-05,
+      "loss": 0.9911,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4877170005326136e-05,
+      "loss": 1.0139,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4874449459114556e-05,
+      "loss": 1.0122,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4871728439607967e-05,
+      "loss": 1.0294,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4869006947070568e-05,
+      "loss": 0.9935,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4866284981766607e-05,
+      "loss": 1.0739,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4863562543960383e-05,
+      "loss": 1.0059,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4860839633916236e-05,
+      "loss": 1.0365,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4858116251898552e-05,
+      "loss": 1.0114,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4855392398171762e-05,
+      "loss": 0.9944,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4852668073000347e-05,
+      "loss": 1.0437,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.484994327664883e-05,
+      "loss": 1.0582,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.484721800938178e-05,
+      "loss": 1.0242,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4844492271463814e-05,
+      "loss": 1.0151,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4841766063159592e-05,
+      "loss": 1.0894,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4839039384733821e-05,
+      "loss": 0.9705,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4836312236451256e-05,
+      "loss": 1.0346,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4833584618576695e-05,
+      "loss": 1.0118,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4830856531374978e-05,
+      "loss": 1.0198,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4828127975111e-05,
+      "loss": 0.9977,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.482539895004969e-05,
+      "loss": 1.0481,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4822669456456031e-05,
+      "loss": 0.9957,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4819939494595052e-05,
+      "loss": 1.0006,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4817209064731819e-05,
+      "loss": 0.9734,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4814478167131454e-05,
+      "loss": 0.9647,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4811746802059115e-05,
+      "loss": 0.979,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4809014969780012e-05,
+      "loss": 1.0022,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.48062826705594e-05,
+      "loss": 0.9882,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4803549904662572e-05,
+      "loss": 0.978,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4800816672354876e-05,
+      "loss": 0.9753,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.47980829739017e-05,
+      "loss": 1.0242,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4795348809568477e-05,
+      "loss": 1.0057,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4792614179620688e-05,
+      "loss": 1.0444,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4789879084323858e-05,
+      "loss": 1.0292,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4787143523943555e-05,
+      "loss": 1.0226,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4784407498745394e-05,
+      "loss": 1.0353,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4781671008995038e-05,
+      "loss": 0.8774,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.477893405495819e-05,
+      "loss": 1.0295,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4776196636900602e-05,
+      "loss": 1.1056,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4773458755088068e-05,
+      "loss": 0.9726,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4770720409786429e-05,
+      "loss": 1.0558,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4767981601261567e-05,
+      "loss": 0.9715,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4765242329779419e-05,
+      "loss": 1.0526,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4762502595605957e-05,
+      "loss": 1.0753,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.47597623990072e-05,
+      "loss": 1.0088,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4757021740249213e-05,
+      "loss": 0.9094,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4754280619598112e-05,
+      "loss": 1.0015,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4751539037320044e-05,
+      "loss": 0.8619,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4748796993681214e-05,
+      "loss": 0.968,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4746054488947863e-05,
+      "loss": 0.9811,
+      "step": 3766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4743311523386283e-05,
+      "loss": 1.0029,
+      "step": 3767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4740568097262811e-05,
+      "loss": 0.9386,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4737824210843816e-05,
+      "loss": 0.9437,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.473507986439573e-05,
+      "loss": 0.9804,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.473233505818502e-05,
+      "loss": 1.0495,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4729589792478193e-05,
+      "loss": 0.9532,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4726844067541817e-05,
+      "loss": 0.9579,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4724097883642482e-05,
+      "loss": 1.0304,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4721351241046844e-05,
+      "loss": 0.9593,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4718604140021588e-05,
+      "loss": 0.9476,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4715856580833454e-05,
+      "loss": 0.9833,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.471310856374922e-05,
+      "loss": 0.9987,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4710360089035709e-05,
+      "loss": 0.9725,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.470761115695979e-05,
+      "loss": 0.9786,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4704861767788384e-05,
+      "loss": 1.0299,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4702111921788437e-05,
+      "loss": 1.0138,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.469936161922696e-05,
+      "loss": 0.9758,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4696610860370997e-05,
+      "loss": 1.0265,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4693859645487635e-05,
+      "loss": 1.026,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4691107974844015e-05,
+      "loss": 1.0267,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.468835584870731e-05,
+      "loss": 1.0771,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.468560326734475e-05,
+      "loss": 0.9843,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.46828502310236e-05,
+      "loss": 0.9017,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4680096740011172e-05,
+      "loss": 1.037,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4677342794574819e-05,
+      "loss": 0.9288,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4674588394981948e-05,
+      "loss": 0.9718,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4671833541499994e-05,
+      "loss": 0.969,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4669078234396454e-05,
+      "loss": 0.9933,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4666322473938857e-05,
+      "loss": 0.9889,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4663566260394775e-05,
+      "loss": 1.0223,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.466080959403184e-05,
+      "loss": 0.9443,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4658052475117704e-05,
+      "loss": 0.9137,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.465529490392008e-05,
+      "loss": 0.9376,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4652536880706723e-05,
+      "loss": 1.0778,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4649778405745422e-05,
+      "loss": 1.0223,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4647019479304028e-05,
+      "loss": 0.96,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4644260101650417e-05,
+      "loss": 0.987,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4641500273052516e-05,
+      "loss": 0.9331,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.46387399937783e-05,
+      "loss": 0.9892,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.463597926409578e-05,
+      "loss": 1.0204,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4633218084273022e-05,
+      "loss": 0.9695,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4630456454578122e-05,
+      "loss": 0.9283,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4627694375279227e-05,
+      "loss": 1.0618,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.462493184664453e-05,
+      "loss": 1.0429,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4622168868942264e-05,
+      "loss": 1.0056,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4619405442440702e-05,
+      "loss": 0.8785,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4616641567408171e-05,
+      "loss": 1.0137,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4613877244113033e-05,
+      "loss": 1.0458,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4611112472823694e-05,
+      "loss": 0.9653,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4608347253808605e-05,
+      "loss": 1.0311,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4605581587336263e-05,
+      "loss": 0.9686,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.460281547367521e-05,
+      "loss": 0.9662,
+      "step": 3818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4600048913094016e-05,
+      "loss": 1.0438,
+      "step": 3819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4597281905861318e-05,
+      "loss": 0.9927,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4594514452245779e-05,
+      "loss": 0.9761,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4591746552516109e-05,
+      "loss": 0.9438,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.458897820694107e-05,
+      "loss": 0.9423,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4586209415789452e-05,
+      "loss": 0.9979,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4583440179330102e-05,
+      "loss": 1.0548,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4580670497831904e-05,
+      "loss": 1.0344,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4577900371563787e-05,
+      "loss": 0.9135,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4575129800794718e-05,
+      "loss": 1.0206,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4572358785793715e-05,
+      "loss": 0.9872,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4569587326829834e-05,
+      "loss": 1.0607,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.456681542417218e-05,
+      "loss": 0.9601,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4564043078089891e-05,
+      "loss": 1.0516,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4561270288852155e-05,
+      "loss": 0.9794,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4558497056728205e-05,
+      "loss": 1.0205,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4555723381987312e-05,
+      "loss": 0.9371,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4552949264898795e-05,
+      "loss": 0.9847,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4550174705732003e-05,
+      "loss": 0.9726,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4547399704756348e-05,
+      "loss": 0.925,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4544624262241269e-05,
+      "loss": 0.9226,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4541848378456255e-05,
+      "loss": 1.0032,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.453907205367084e-05,
+      "loss": 0.9751,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4536295288154594e-05,
+      "loss": 0.9334,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4533518082177129e-05,
+      "loss": 0.912,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4530740436008111e-05,
+      "loss": 1.0289,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4527962349917234e-05,
+      "loss": 0.9829,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.452518382417425e-05,
+      "loss": 0.9462,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4522404859048939e-05,
+      "loss": 0.9061,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4519625454811135e-05,
+      "loss": 0.9885,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4516845611730707e-05,
+      "loss": 1.0496,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4514065330077575e-05,
+      "loss": 0.9963,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4511284610121688e-05,
+      "loss": 0.9321,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4508503452133053e-05,
+      "loss": 1.006,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.450572185638171e-05,
+      "loss": 1.0015,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4502939823137744e-05,
+      "loss": 1.019,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4500157352671285e-05,
+      "loss": 0.9247,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4497374445252496e-05,
+      "loss": 1.0139,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4494591101151594e-05,
+      "loss": 1.0257,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4491807320638835e-05,
+      "loss": 0.9911,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4489023103984513e-05,
+      "loss": 1.0454,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4486238451458972e-05,
+      "loss": 0.9947,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4483453363332586e-05,
+      "loss": 0.9548,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4480667839875786e-05,
+      "loss": 1.0072,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4477881881359034e-05,
+      "loss": 0.9282,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4475095488052843e-05,
+      "loss": 1.0175,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4472308660227761e-05,
+      "loss": 0.9151,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4469521398154381e-05,
+      "loss": 0.994,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4466733702103337e-05,
+      "loss": 1.0208,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4463945572345308e-05,
+      "loss": 1.0141,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4461157009151012e-05,
+      "loss": 0.9985,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4458368012791213e-05,
+      "loss": 1.0458,
+      "step": 3870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4455578583536714e-05,
+      "loss": 0.8981,
+      "step": 3871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4452788721658355e-05,
+      "loss": 0.9702,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.444999842742703e-05,
+      "loss": 0.9292,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4447207701113669e-05,
+      "loss": 0.9447,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.444441654298924e-05,
+      "loss": 0.9449,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4441624953324755e-05,
+      "loss": 1.0112,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4438832932391273e-05,
+      "loss": 1.001,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4436040480459891e-05,
+      "loss": 0.9953,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4433247597801749e-05,
+      "loss": 0.9134,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.443045428468802e-05,
+      "loss": 1.0252,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4427660541389939e-05,
+      "loss": 0.9681,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4424866368178761e-05,
+      "loss": 0.913,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4422071765325794e-05,
+      "loss": 0.9698,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.441927673310239e-05,
+      "loss": 1.0283,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4416481271779931e-05,
+      "loss": 1.0385,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4413685381629855e-05,
+      "loss": 0.9734,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4410889062923634e-05,
+      "loss": 0.9067,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.440809231593278e-05,
+      "loss": 1.0515,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4405295140928853e-05,
+      "loss": 1.005,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4402497538183444e-05,
+      "loss": 0.9911,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4399699507968199e-05,
+      "loss": 0.9523,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4396901050554794e-05,
+      "loss": 1.0096,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4394102166214955e-05,
+      "loss": 0.9962,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4391302855220442e-05,
+      "loss": 1.0068,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4388503117843062e-05,
+      "loss": 1.0223,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4385702954354662e-05,
+      "loss": 0.9708,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4382902365027129e-05,
+      "loss": 1.0011,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.438010135013239e-05,
+      "loss": 0.9692,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4377299909942417e-05,
+      "loss": 1.0781,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4374498044729225e-05,
+      "loss": 1.0718,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4371695754764865e-05,
+      "loss": 0.8579,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4368893040321428e-05,
+      "loss": 1.0144,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4366089901671054e-05,
+      "loss": 0.978,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4363286339085915e-05,
+      "loss": 1.0137,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4360482352838237e-05,
+      "loss": 0.9349,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.435767794320027e-05,
+      "loss": 0.9875,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.435487311044432e-05,
+      "loss": 0.9936,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4352067854842724e-05,
+      "loss": 0.9706,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4349262176667863e-05,
+      "loss": 0.9859,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.434645607619217e-05,
+      "loss": 0.9463,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4343649553688099e-05,
+      "loss": 0.9863,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.434084260942816e-05,
+      "loss": 1.0862,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.43380352436849e-05,
+      "loss": 0.9744,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4335227456730902e-05,
+      "loss": 0.9354,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4332419248838799e-05,
+      "loss": 0.9745,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4329610620281253e-05,
+      "loss": 0.9929,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4326801571330981e-05,
+      "loss": 0.9894,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4323992102260733e-05,
+      "loss": 1.0226,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4321182213343295e-05,
+      "loss": 0.9993,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4318371904851502e-05,
+      "loss": 0.9472,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4315561177058227e-05,
+      "loss": 0.9816,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4312750030236382e-05,
+      "loss": 1.0007,
+      "step": 3922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4309938464658926e-05,
+      "loss": 0.9852,
+      "step": 3923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4307126480598852e-05,
+      "loss": 0.9784,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4304314078329192e-05,
+      "loss": 1.0618,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4301501258123024e-05,
+      "loss": 1.0294,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4298688020253464e-05,
+      "loss": 0.9522,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4295874364993672e-05,
+      "loss": 1.0219,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4293060292616847e-05,
+      "loss": 1.035,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4290245803396221e-05,
+      "loss": 0.9062,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4287430897605078e-05,
+      "loss": 1.0752,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4284615575516737e-05,
+      "loss": 0.9756,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4281799837404553e-05,
+      "loss": 0.9714,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4278983683541934e-05,
+      "loss": 1.0787,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4276167114202314e-05,
+      "loss": 1.0117,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4273350129659173e-05,
+      "loss": 0.9883,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4270532730186037e-05,
+      "loss": 0.9756,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4267714916056465e-05,
+      "loss": 1.0096,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4264896687544059e-05,
+      "loss": 1.0067,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.426207804492246e-05,
+      "loss": 1.0386,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4259258988465351e-05,
+      "loss": 1.0089,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4256439518446456e-05,
+      "loss": 0.929,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4253619635139535e-05,
+      "loss": 1.0367,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4250799338818388e-05,
+      "loss": 0.9272,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4247978629756865e-05,
+      "loss": 1.017,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.424515750822884e-05,
+      "loss": 0.9694,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4242335974508243e-05,
+      "loss": 1.0497,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4239514028869032e-05,
+      "loss": 1.0299,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4236691671585215e-05,
+      "loss": 0.9542,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4233868902930827e-05,
+      "loss": 1.0062,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4231045723179959e-05,
+      "loss": 1.0397,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4228222132606729e-05,
+      "loss": 0.9938,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.42253981314853e-05,
+      "loss": 0.9789,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4222573720089874e-05,
+      "loss": 1.0134,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4219748898694698e-05,
+      "loss": 1.0237,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4216923667574042e-05,
+      "loss": 1.0104,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4214098027002243e-05,
+      "loss": 1.009,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4211271977253653e-05,
+      "loss": 0.9799,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4208445518602675e-05,
+      "loss": 0.957,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4205618651323753e-05,
+      "loss": 1.1146,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4202791375691365e-05,
+      "loss": 0.9553,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4199963691980027e-05,
+      "loss": 1.0334,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4197135600464312e-05,
+      "loss": 1.0546,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4194307101418805e-05,
+      "loss": 1.0082,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4191478195118152e-05,
+      "loss": 0.9153,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4188648881837033e-05,
+      "loss": 0.8941,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4185819161850162e-05,
+      "loss": 1.0106,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4182989035432299e-05,
+      "loss": 0.9443,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4180158502858242e-05,
+      "loss": 0.9763,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4177327564402825e-05,
+      "loss": 0.9783,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.417449622034093e-05,
+      "loss": 1.0255,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4171664470947464e-05,
+      "loss": 0.9521,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4168832316497386e-05,
+      "loss": 1.0331,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.416599975726569e-05,
+      "loss": 0.9442,
+      "step": 3974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4163166793527408e-05,
+      "loss": 0.9957,
+      "step": 3975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4160333425557616e-05,
+      "loss": 1.0664,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4157499653631419e-05,
+      "loss": 1.0098,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4154665478023977e-05,
+      "loss": 1.042,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4151830899010473e-05,
+      "loss": 1.0316,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4148995916866139e-05,
+      "loss": 0.9946,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4146160531866247e-05,
+      "loss": 1.0562,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.41433247442861e-05,
+      "loss": 1.0231,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4140488554401047e-05,
+      "loss": 0.9778,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4137651962486472e-05,
+      "loss": 0.9687,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.41348149688178e-05,
+      "loss": 1.0322,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4131977573670499e-05,
+      "loss": 0.9543,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4129139777320069e-05,
+      "loss": 0.9354,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.412630158004205e-05,
+      "loss": 0.9947,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4123462982112026e-05,
+      "loss": 0.9745,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4120623983805617e-05,
+      "loss": 0.9642,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4117784585398475e-05,
+      "loss": 1.002,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4114944787166307e-05,
+      "loss": 1.0486,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4112104589384843e-05,
+      "loss": 0.9586,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4109263992329858e-05,
+      "loss": 1.0558,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4106422996277172e-05,
+      "loss": 1.0085,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4103581601502629e-05,
+      "loss": 0.9471,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4100739808282124e-05,
+      "loss": 1.0402,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.409789761689159e-05,
+      "loss": 0.9936,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.409505502760699e-05,
+      "loss": 1.0863,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4092212040704336e-05,
+      "loss": 0.924,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.408936865645967e-05,
+      "loss": 0.9671,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.408652487514908e-05,
+      "loss": 1.034,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4083680697048685e-05,
+      "loss": 0.9838,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.408083612243465e-05,
+      "loss": 0.9768,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4077991151583175e-05,
+      "loss": 0.8935,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4075145784770496e-05,
+      "loss": 0.9567,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4072300022272891e-05,
+      "loss": 1.03,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4069453864366678e-05,
+      "loss": 0.9618,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4066607311328204e-05,
+      "loss": 1.1201,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4063760363433867e-05,
+      "loss": 0.9929,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4060913020960096e-05,
+      "loss": 0.963,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.405806528418336e-05,
+      "loss": 1.0369,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4055217153380167e-05,
+      "loss": 0.8946,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4052368628827057e-05,
+      "loss": 1.0246,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.404951971080062e-05,
+      "loss": 0.9026,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4046670399577478e-05,
+      "loss": 1.0352,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4043820695434284e-05,
+      "loss": 1.0739,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4040970598647742e-05,
+      "loss": 0.9974,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4038120109494587e-05,
+      "loss": 1.0119,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4035269228251589e-05,
+      "loss": 1.03,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4032417955195568e-05,
+      "loss": 1.0292,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4029566290603368e-05,
+      "loss": 1.05,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4026714234751879e-05,
+      "loss": 0.9648,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4023861787918031e-05,
+      "loss": 1.0024,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4021008950378785e-05,
+      "loss": 0.9616,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4018155722411144e-05,
+      "loss": 0.9566,
+      "step": 4026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4015302104292148e-05,
+      "loss": 1.0582,
+      "step": 4027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4012448096298874e-05,
+      "loss": 0.9208,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4009593698708442e-05,
+      "loss": 1.0149,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4006738911798001e-05,
+      "loss": 0.8988,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4003883735844743e-05,
+      "loss": 0.8947,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.40010281711259e-05,
+      "loss": 0.9461,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3998172217918739e-05,
+      "loss": 1.09,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3995315876500565e-05,
+      "loss": 1.0905,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.399245914714872e-05,
+      "loss": 0.9745,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3989602030140581e-05,
+      "loss": 0.9706,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.398674452575357e-05,
+      "loss": 0.9362,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.398388663426514e-05,
+      "loss": 1.0188,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3981028355952786e-05,
+      "loss": 1.0085,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3978169691094037e-05,
+      "loss": 1.0239,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3975310639966462e-05,
+      "loss": 1.0347,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3972451202847665e-05,
+      "loss": 0.9301,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3969591380015295e-05,
+      "loss": 0.9644,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3966731171747024e-05,
+      "loss": 0.9495,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3963870578320576e-05,
+      "loss": 1.0212,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3961009600013702e-05,
+      "loss": 0.9437,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3958148237104196e-05,
+      "loss": 0.9845,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3955286489869894e-05,
+      "loss": 1.0266,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3952424358588655e-05,
+      "loss": 1.0005,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.394956184353839e-05,
+      "loss": 0.9597,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3946698944997037e-05,
+      "loss": 0.9663,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3943835663242577e-05,
+      "loss": 0.9388,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3940971998553028e-05,
+      "loss": 1.05,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3938107951206438e-05,
+      "loss": 0.9647,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3935243521480902e-05,
+      "loss": 1.1204,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3932378709654548e-05,
+      "loss": 1.0045,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.392951351600554e-05,
+      "loss": 0.8949,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3926647940812081e-05,
+      "loss": 0.9994,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.392378198435241e-05,
+      "loss": 0.9356,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.39209156469048e-05,
+      "loss": 0.9124,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.391804892874757e-05,
+      "loss": 0.9625,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3915181830159061e-05,
+      "loss": 0.9674,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3912314351417673e-05,
+      "loss": 1.0422,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3909446492801819e-05,
+      "loss": 0.9784,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3906578254589962e-05,
+      "loss": 1.0219,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3903709637060605e-05,
+      "loss": 1.0076,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3900840640492274e-05,
+      "loss": 0.9666,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3897971265163546e-05,
+      "loss": 0.9763,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3895101511353031e-05,
+      "loss": 1.0351,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3892231379339369e-05,
+      "loss": 0.9869,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3889360869401243e-05,
+      "loss": 0.9412,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3886489981817375e-05,
+      "loss": 0.9813,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3883618716866512e-05,
+      "loss": 0.976,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3880747074827454e-05,
+      "loss": 0.9662,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3877875055979021e-05,
+      "loss": 0.9918,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3875002660600085e-05,
+      "loss": 0.9672,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3872129888969545e-05,
+      "loss": 1.0577,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.386925674136634e-05,
+      "loss": 1.0583,
+      "step": 4078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3866383218069439e-05,
+      "loss": 1.0787,
+      "step": 4079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3863509319357857e-05,
+      "loss": 1.0439,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.386063504551064e-05,
+      "loss": 0.9944,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3857760396806876e-05,
+      "loss": 1.0131,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3854885373525679e-05,
+      "loss": 0.9973,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3852009975946209e-05,
+      "loss": 0.9435,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3849134204347656e-05,
+      "loss": 1.0141,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3846258059009252e-05,
+      "loss": 0.9508,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3843381540210263e-05,
+      "loss": 1.0984,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.384050464822999e-05,
+      "loss": 1.0311,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3837627383347767e-05,
+      "loss": 0.9484,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.383474974584297e-05,
+      "loss": 0.938,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3831871735995015e-05,
+      "loss": 1.0013,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3828993354083342e-05,
+      "loss": 1.0095,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3826114600387437e-05,
+      "loss": 1.0281,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3823235475186816e-05,
+      "loss": 0.9618,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3820355978761037e-05,
+      "loss": 0.9435,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3817476111389685e-05,
+      "loss": 1.0161,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3814595873352392e-05,
+      "loss": 0.7921,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3811715264928824e-05,
+      "loss": 0.9329,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3808834286398672e-05,
+      "loss": 1.0577,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3805952938041674e-05,
+      "loss": 0.9522,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3803071220137602e-05,
+      "loss": 0.9703,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3800189132966257e-05,
+      "loss": 1.0581,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3797306676807491e-05,
+      "loss": 0.9705,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3794423851941174e-05,
+      "loss": 0.9858,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3791540658647224e-05,
+      "loss": 1.0976,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.378865709720559e-05,
+      "loss": 1.0167,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3785773167896257e-05,
+      "loss": 1.0258,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3782888870999245e-05,
+      "loss": 1.0135,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3780004206794613e-05,
+      "loss": 0.9523,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.377711917556245e-05,
+      "loss": 1.0104,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3774233777582893e-05,
+      "loss": 0.9849,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3771348013136096e-05,
+      "loss": 0.9527,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3768461882502263e-05,
+      "loss": 0.9841,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3765575385961627e-05,
+      "loss": 0.9232,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3762688523794461e-05,
+      "loss": 1.0008,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3759801296281072e-05,
+      "loss": 0.9685,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3756913703701797e-05,
+      "loss": 0.9625,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3754025746337014e-05,
+      "loss": 1.0272,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.375113742446714e-05,
+      "loss": 1.0055,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3748248738372616e-05,
+      "loss": 0.9567,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.374535968833393e-05,
+      "loss": 0.9995,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3742470274631599e-05,
+      "loss": 0.9171,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3739580497546177e-05,
+      "loss": 0.9341,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3736690357358253e-05,
+      "loss": 1.0765,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3733799854348448e-05,
+      "loss": 0.9612,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3730908988797427e-05,
+      "loss": 1.0327,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3728017760985883e-05,
+      "loss": 0.9202,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3725126171194543e-05,
+      "loss": 0.9558,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3722234219704175e-05,
+      "loss": 0.9074,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.371934190679558e-05,
+      "loss": 1.0717,
+      "step": 4130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3716449232749587e-05,
+      "loss": 1.0411,
+      "step": 4131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3713556197847076e-05,
+      "loss": 0.9136,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3710662802368946e-05,
+      "loss": 0.9808,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3707769046596136e-05,
+      "loss": 1.0644,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3704874930809626e-05,
+      "loss": 0.9398,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3701980455290425e-05,
+      "loss": 0.9585,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3699085620319576e-05,
+      "loss": 1.0441,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3696190426178162e-05,
+      "loss": 1.0596,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3693294873147297e-05,
+      "loss": 0.9543,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3690398961508128e-05,
+      "loss": 0.9027,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3687502691541843e-05,
+      "loss": 0.9448,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3684606063529662e-05,
+      "loss": 0.9946,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3681709077752838e-05,
+      "loss": 1.0508,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3678811734492659e-05,
+      "loss": 0.881,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.367591403403045e-05,
+      "loss": 0.9963,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.367301597664757e-05,
+      "loss": 0.9787,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3670117562625409e-05,
+      "loss": 0.9582,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.36672187922454e-05,
+      "loss": 1.0862,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3664319665788997e-05,
+      "loss": 1.0509,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3661420183537705e-05,
+      "loss": 0.9263,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3658520345773055e-05,
+      "loss": 0.949,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3655620152776605e-05,
+      "loss": 1.0243,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3652719604829962e-05,
+      "loss": 1.0014,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.364981870221476e-05,
+      "loss": 1.0613,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3646917445212666e-05,
+      "loss": 1.0114,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.364401583410539e-05,
+      "loss": 0.957,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.364111386917466e-05,
+      "loss": 1.0078,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3638211550702256e-05,
+      "loss": 1.0149,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3635308878969984e-05,
+      "loss": 1.0053,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.363240585425968e-05,
+      "loss": 0.9481,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3629502476853226e-05,
+      "loss": 1.0029,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.362659874703253e-05,
+      "loss": 1.0041,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.362369466507953e-05,
+      "loss": 0.9977,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3620790231276213e-05,
+      "loss": 0.988,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3617885445904584e-05,
+      "loss": 0.9144,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3614980309246692e-05,
+      "loss": 1.027,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3612074821584615e-05,
+      "loss": 1.0051,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3609168983200474e-05,
+      "loss": 0.8415,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3606262794376412e-05,
+      "loss": 0.993,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3603356255394613e-05,
+      "loss": 1.022,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3600449366537294e-05,
+      "loss": 1.1144,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3597542128086702e-05,
+      "loss": 1.0186,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3594634540325127e-05,
+      "loss": 1.0477,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3591726603534885e-05,
+      "loss": 0.9472,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3588818317998329e-05,
+      "loss": 0.9919,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3585909683997842e-05,
+      "loss": 0.9508,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3583000701815849e-05,
+      "loss": 1.0186,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3580091371734798e-05,
+      "loss": 1.0557,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.357718169403718e-05,
+      "loss": 1.033,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.357427166900552e-05,
+      "loss": 0.9412,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3571361296922365e-05,
+      "loss": 0.9644,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3568450578070309e-05,
+      "loss": 0.9553,
+      "step": 4182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3565539512731976e-05,
+      "loss": 0.9639,
+      "step": 4183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3562628101190015e-05,
+      "loss": 0.9372,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3559716343727122e-05,
+      "loss": 0.8894,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3556804240626019e-05,
+      "loss": 0.995,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.355389179216946e-05,
+      "loss": 0.9806,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3550978998640241e-05,
+      "loss": 0.989,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.354806586032118e-05,
+      "loss": 0.9751,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3545152377495136e-05,
+      "loss": 0.9859,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3542238550444999e-05,
+      "loss": 1.0024,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3539324379453698e-05,
+      "loss": 0.9235,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3536409864804187e-05,
+      "loss": 1.0238,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3533495006779455e-05,
+      "loss": 1.0183,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3530579805662531e-05,
+      "loss": 0.9954,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3527664261736471e-05,
+      "loss": 0.9857,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3524748375284362e-05,
+      "loss": 0.9765,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3521832146589335e-05,
+      "loss": 0.9949,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3518915575934541e-05,
+      "loss": 0.9332,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3515998663603174e-05,
+      "loss": 0.9933,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3513081409878456e-05,
+      "loss": 0.921,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3510163815043647e-05,
+      "loss": 0.9657,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3507245879382037e-05,
+      "loss": 1.0222,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3504327603176943e-05,
+      "loss": 1.0286,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3501408986711729e-05,
+      "loss": 0.9985,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3498490030269782e-05,
+      "loss": 0.9976,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3495570734134517e-05,
+      "loss": 0.9826,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3492651098589398e-05,
+      "loss": 1.0232,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3489731123917914e-05,
+      "loss": 0.9564,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3486810810403578e-05,
+      "loss": 1.0262,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.348389015832995e-05,
+      "loss": 0.956,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.348096916798062e-05,
+      "loss": 0.9816,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3478047839639197e-05,
+      "loss": 0.9702,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3475126173589343e-05,
+      "loss": 0.962,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3472204170114742e-05,
+      "loss": 1.0521,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3469281829499107e-05,
+      "loss": 0.9242,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3466359152026197e-05,
+      "loss": 1.0662,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3463436137979786e-05,
+      "loss": 1.0272,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3460512787643695e-05,
+      "loss": 1.0313,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3457589101301776e-05,
+      "loss": 0.9957,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3454665079237907e-05,
+      "loss": 1.0252,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3451740721736005e-05,
+      "loss": 0.9969,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.344881602908001e-05,
+      "loss": 0.9811,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3445891001553905e-05,
+      "loss": 0.9877,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3442965639441705e-05,
+      "loss": 0.9233,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3440039943027452e-05,
+      "loss": 0.992,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3437113912595223e-05,
+      "loss": 0.9871,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3434187548429126e-05,
+      "loss": 1.0545,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.34312608508133e-05,
+      "loss": 0.9839,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3428333820031922e-05,
+      "loss": 1.0389,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.34254064563692e-05,
+      "loss": 0.9223,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3422478760109371e-05,
+      "loss": 0.9805,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3419550731536708e-05,
+      "loss": 0.9751,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3416622370935507e-05,
+      "loss": 1.0468,
+      "step": 4234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3413693678590111e-05,
+      "loss": 0.8906,
+      "step": 4235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3410764654784885e-05,
+      "loss": 0.9777,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3407835299804226e-05,
+      "loss": 0.9498,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3404905613932573e-05,
+      "loss": 1.0419,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3401975597454379e-05,
+      "loss": 1.0524,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3399045250654152e-05,
+      "loss": 0.9399,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3396114573816414e-05,
+      "loss": 0.9419,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3393183567225724e-05,
+      "loss": 0.9045,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3390252231166679e-05,
+      "loss": 0.9342,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3387320565923901e-05,
+      "loss": 0.8645,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3384388571782043e-05,
+      "loss": 1.0409,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.33814562490258e-05,
+      "loss": 0.8826,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3378523597939887e-05,
+      "loss": 0.9948,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3375590618809056e-05,
+      "loss": 0.9823,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3372657311918093e-05,
+      "loss": 0.9011,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3369723677551813e-05,
+      "loss": 0.9577,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3366789715995065e-05,
+      "loss": 1.0266,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3363855427532724e-05,
+      "loss": 0.9688,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3360920812449704e-05,
+      "loss": 1.0145,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3357985871030948e-05,
+      "loss": 1.0245,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3355050603561428e-05,
+      "loss": 0.9639,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3352115010326155e-05,
+      "loss": 1.0195,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3349179091610159e-05,
+      "loss": 0.986,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3346242847698516e-05,
+      "loss": 0.9971,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3343306278876324e-05,
+      "loss": 0.9233,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3340369385428713e-05,
+      "loss": 1.0252,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3337432167640854e-05,
+      "loss": 1.0714,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3334494625797936e-05,
+      "loss": 0.9701,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3331556760185186e-05,
+      "loss": 0.9619,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3328618571087867e-05,
+      "loss": 0.9672,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3325680058791263e-05,
+      "loss": 0.9728,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.33227412235807e-05,
+      "loss": 1.0061,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3319802065741527e-05,
+      "loss": 1.0023,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3316862585559132e-05,
+      "loss": 1.0676,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3313922783318922e-05,
+      "loss": 0.9814,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3310982659306352e-05,
+      "loss": 0.9907,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3308042213806895e-05,
+      "loss": 1.0212,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3305101447106064e-05,
+      "loss": 1.0668,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.330216035948939e-05,
+      "loss": 0.9777,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3299218951242456e-05,
+      "loss": 0.9246,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3296277222650856e-05,
+      "loss": 1.0035,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3293335174000226e-05,
+      "loss": 0.9975,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.329039280557623e-05,
+      "loss": 0.9996,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.328745011766456e-05,
+      "loss": 0.9755,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3284507110550951e-05,
+      "loss": 0.9548,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3281563784521154e-05,
+      "loss": 1.0111,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3278620139860959e-05,
+      "loss": 1.0225,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3275676176856185e-05,
+      "loss": 0.9618,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3272731895792682e-05,
+      "loss": 0.9245,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3269787296956333e-05,
+      "loss": 1.025,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3266842380633054e-05,
+      "loss": 0.9683,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3263897147108778e-05,
+      "loss": 0.9586,
+      "step": 4286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3260951596669483e-05,
+      "loss": 1.0042,
+      "step": 4287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3258005729601178e-05,
+      "loss": 0.9807,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.325505954618989e-05,
+      "loss": 1.0404,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3252113046721692e-05,
+      "loss": 0.9863,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.324916623148268e-05,
+      "loss": 1.0148,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3246219100758974e-05,
+      "loss": 1.034,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3243271654836745e-05,
+      "loss": 0.9715,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3240323894002166e-05,
+      "loss": 1.0092,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3237375818541467e-05,
+      "loss": 0.9324,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3234427428740895e-05,
+      "loss": 1.0159,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3231478724886729e-05,
+      "loss": 1.0316,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3228529707265279e-05,
+      "loss": 0.9832,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3225580376162891e-05,
+      "loss": 1.04,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.322263073186593e-05,
+      "loss": 0.9685,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.32196807746608e-05,
+      "loss": 0.9762,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3216730504833938e-05,
+      "loss": 1.0472,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3213779922671798e-05,
+      "loss": 0.9279,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3210829028460883e-05,
+      "loss": 0.917,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.320787782248771e-05,
+      "loss": 0.9983,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3204926305038832e-05,
+      "loss": 0.9753,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3201974476400836e-05,
+      "loss": 1.0664,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3199022336860335e-05,
+      "loss": 0.9941,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3196069886703975e-05,
+      "loss": 1.0171,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3193117126218425e-05,
+      "loss": 0.9721,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3190164055690393e-05,
+      "loss": 0.8931,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3187210675406617e-05,
+      "loss": 1.0067,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3184256985653856e-05,
+      "loss": 1.0055,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.318130298671891e-05,
+      "loss": 1.0272,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.31783486788886e-05,
+      "loss": 1.1158,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3175394062449777e-05,
+      "loss": 1.0107,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3172439137689339e-05,
+      "loss": 0.9835,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3169483904894185e-05,
+      "loss": 0.9766,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.316652836435127e-05,
+      "loss": 0.9112,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3163572516347565e-05,
+      "loss": 1.0474,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3160616361170076e-05,
+      "loss": 0.9708,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3157659899105835e-05,
+      "loss": 0.9677,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3154703130441906e-05,
+      "loss": 0.9178,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.315174605546538e-05,
+      "loss": 1.0198,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.314878867446339e-05,
+      "loss": 1.0598,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3145830987723081e-05,
+      "loss": 1.0106,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3142872995531637e-05,
+      "loss": 1.0841,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3139914698176273e-05,
+      "loss": 1.0084,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.313695609594423e-05,
+      "loss": 1.024,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3133997189122777e-05,
+      "loss": 1.0188,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.313103797799922e-05,
+      "loss": 0.9542,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3128078462860887e-05,
+      "loss": 0.9376,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.312511864399514e-05,
+      "loss": 0.9136,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3122158521689367e-05,
+      "loss": 0.9787,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3119198096230989e-05,
+      "loss": 1.0047,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3116237367907454e-05,
+      "loss": 1.0756,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3113276337006237e-05,
+      "loss": 0.9005,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3110315003814855e-05,
+      "loss": 0.9734,
+      "step": 4338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3107353368620835e-05,
+      "loss": 1.0319,
+      "step": 4339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3104391431711748e-05,
+      "loss": 0.9728,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101429193375188e-05,
+      "loss": 0.9388,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309846665389878e-05,
+      "loss": 0.9835,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3095503813570182e-05,
+      "loss": 1.0229,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309254067267707e-05,
+      "loss": 1.0004,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3089577231507162e-05,
+      "loss": 1.0226,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3086613490348198e-05,
+      "loss": 1.0248,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3083649449487946e-05,
+      "loss": 0.8547,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3080685109214208e-05,
+      "loss": 0.9866,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3077720469814818e-05,
+      "loss": 0.9565,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3074755531577628e-05,
+      "loss": 0.9515,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3071790294790524e-05,
+      "loss": 1.0007,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3068824759741428e-05,
+      "loss": 1.0657,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3065858926718276e-05,
+      "loss": 0.9891,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306289279600905e-05,
+      "loss": 0.9844,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3059926367901747e-05,
+      "loss": 0.9645,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056959642684404e-05,
+      "loss": 0.9594,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305399262064508e-05,
+      "loss": 0.9981,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305102530207186e-05,
+      "loss": 1.0094,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3048057687252866e-05,
+      "loss": 0.9268,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3045089776476246e-05,
+      "loss": 0.9477,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304212157003017e-05,
+      "loss": 1.015,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3039153068202853e-05,
+      "loss": 1.0628,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3036184271282515e-05,
+      "loss": 1.0121,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3033215179557424e-05,
+      "loss": 0.9944,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303024579331587e-05,
+      "loss": 1.0051,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3027276112846172e-05,
+      "loss": 1.0199,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.302430613843668e-05,
+      "loss": 1.0496,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3021335870375763e-05,
+      "loss": 0.9475,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3018365308951832e-05,
+      "loss": 1.0289,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3015394454453316e-05,
+      "loss": 1.088,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3012423307168677e-05,
+      "loss": 1.0958,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3009451867386411e-05,
+      "loss": 0.9546,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3006480135395027e-05,
+      "loss": 0.995,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3003508111483077e-05,
+      "loss": 0.8727,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3000535795939135e-05,
+      "loss": 0.9402,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.29975631890518e-05,
+      "loss": 1.0247,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299459029110971e-05,
+      "loss": 0.9453,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2991617102401524e-05,
+      "loss": 0.9249,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2988643623215926e-05,
+      "loss": 1.0149,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2985669853841635e-05,
+      "loss": 1.0254,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.29826957945674e-05,
+      "loss": 0.9214,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297972144568198e-05,
+      "loss": 1.0225,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2976746807474193e-05,
+      "loss": 0.9868,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2973771880232853e-05,
+      "loss": 0.9932,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2970796664246826e-05,
+      "loss": 1.0177,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2967821159804994e-05,
+      "loss": 0.9821,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.296484536719627e-05,
+      "loss": 1.0327,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2961869286709594e-05,
+      "loss": 0.9527,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958892918633934e-05,
+      "loss": 0.9835,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295591626325829e-05,
+      "loss": 0.9927,
+      "step": 4390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2952939320871687e-05,
+      "loss": 1.0016,
+      "step": 4391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2949962091763174e-05,
+      "loss": 1.0125,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2946984576221833e-05,
+      "loss": 1.0292,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2944006774536773e-05,
+      "loss": 1.0134,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2941028686997128e-05,
+      "loss": 1.0081,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2938050313892062e-05,
+      "loss": 0.9594,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2935071655510771e-05,
+      "loss": 1.0129,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2932092712142468e-05,
+      "loss": 0.9565,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2929113484076404e-05,
+      "loss": 1.0178,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.292613397160185e-05,
+      "loss": 0.9559,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2923154175008111e-05,
+      "loss": 0.955,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2920174094584514e-05,
+      "loss": 0.9612,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2917193730620418e-05,
+      "loss": 1.0424,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2914213083405211e-05,
+      "loss": 0.9879,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2911232153228298e-05,
+      "loss": 1.0785,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2908250940379124e-05,
+      "loss": 0.9456,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2905269445147154e-05,
+      "loss": 0.9885,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2902287667821885e-05,
+      "loss": 0.9371,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2899305608692835e-05,
+      "loss": 1.0342,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.289632326804956e-05,
+      "loss": 0.9503,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2893340646181627e-05,
+      "loss": 0.8964,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2890357743378649e-05,
+      "loss": 0.9848,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2887374559930254e-05,
+      "loss": 1.0336,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2884391096126098e-05,
+      "loss": 0.9031,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2881407352255872e-05,
+      "loss": 0.9398,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2878423328609281e-05,
+      "loss": 0.9894,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2875439025476075e-05,
+      "loss": 1.0366,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2872454443146015e-05,
+      "loss": 0.9655,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2869469581908893e-05,
+      "loss": 1.013,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.286648444205454e-05,
+      "loss": 0.944,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2863499023872794e-05,
+      "loss": 0.9875,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2860513327653537e-05,
+      "loss": 0.9209,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.285752735368667e-05,
+      "loss": 0.9971,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2854541102262119e-05,
+      "loss": 0.9924,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2851554573669847e-05,
+      "loss": 0.9854,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.284856776819983e-05,
+      "loss": 1.0612,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2845580686142088e-05,
+      "loss": 0.936,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2842593327786649e-05,
+      "loss": 0.9214,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2839605693423575e-05,
+      "loss": 1.0169,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2836617783342968e-05,
+      "loss": 1.0507,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2833629597834937e-05,
+      "loss": 0.9536,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2830641137189628e-05,
+      "loss": 0.9393,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2827652401697214e-05,
+      "loss": 0.9772,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.282466339164789e-05,
+      "loss": 0.9341,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2821674107331878e-05,
+      "loss": 0.9716,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2818684549039437e-05,
+      "loss": 0.9969,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2815694717060837e-05,
+      "loss": 1.0693,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2812704611686386e-05,
+      "loss": 1.0018,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2809714233206414e-05,
+      "loss": 1.0231,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2806723581911274e-05,
+      "loss": 1.006,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2803732658091357e-05,
+      "loss": 0.9499,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2800741462037065e-05,
+      "loss": 0.9888,
+      "step": 4442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2797749994038841e-05,
+      "loss": 1.0052,
+      "step": 4443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2794758254387147e-05,
+      "loss": 0.957,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.279176624337247e-05,
+      "loss": 0.909,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2788773961285323e-05,
+      "loss": 1.0165,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2785781408416256e-05,
+      "loss": 0.9645,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2782788585055829e-05,
+      "loss": 0.9968,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2779795491494644e-05,
+      "loss": 1.0118,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2776802128023317e-05,
+      "loss": 0.9693,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2773808494932496e-05,
+      "loss": 0.9772,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2770814592512853e-05,
+      "loss": 1.0245,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2767820421055089e-05,
+      "loss": 0.9608,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2764825980849931e-05,
+      "loss": 0.9446,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2761831272188127e-05,
+      "loss": 0.9912,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2758836295360455e-05,
+      "loss": 0.935,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2755841050657725e-05,
+      "loss": 0.9758,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2752845538370752e-05,
+      "loss": 0.9636,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2749849758790407e-05,
+      "loss": 0.923,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2746853712207567e-05,
+      "loss": 0.9608,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2743857398913135e-05,
+      "loss": 0.9753,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.274086081919805e-05,
+      "loss": 1.0037,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2737863973353265e-05,
+      "loss": 1.0024,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.273486686166977e-05,
+      "loss": 0.9698,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2731869484438577e-05,
+      "loss": 1.0093,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2728871841950719e-05,
+      "loss": 0.9631,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.272587393449726e-05,
+      "loss": 0.9366,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2722875762369288e-05,
+      "loss": 1.0708,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2719877325857919e-05,
+      "loss": 1.0259,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2716878625254287e-05,
+      "loss": 0.9934,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2713879660849562e-05,
+      "loss": 0.9507,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2710880432934934e-05,
+      "loss": 0.9846,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.270788094180162e-05,
+      "loss": 1.0522,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.270488118774086e-05,
+      "loss": 0.936,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2701881171043922e-05,
+      "loss": 0.9812,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.26988808920021e-05,
+      "loss": 1.041,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.269588035090671e-05,
+      "loss": 1.0573,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.26928795480491e-05,
+      "loss": 1.0558,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2689878483720636e-05,
+      "loss": 0.9464,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2686877158212715e-05,
+      "loss": 1.03,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2683875571816753e-05,
+      "loss": 0.9762,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.26808737248242e-05,
+      "loss": 0.9539,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2677871617526524e-05,
+      "loss": 0.9323,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2674869250215225e-05,
+      "loss": 0.9333,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.267186662318182e-05,
+      "loss": 1.0146,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2668863736717855e-05,
+      "loss": 0.8999,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2665860591114901e-05,
+      "loss": 0.927,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2662857186664558e-05,
+      "loss": 0.941,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2659853523658452e-05,
+      "loss": 1.0075,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2656849602388222e-05,
+      "loss": 0.9571,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2653845423145543e-05,
+      "loss": 0.8842,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2650840986222111e-05,
+      "loss": 1.0376,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2647836291909647e-05,
+      "loss": 0.9931,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2644831340499906e-05,
+      "loss": 0.9283,
+      "step": 4494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2641826132284652e-05,
+      "loss": 1.0242,
+      "step": 4495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2638820667555685e-05,
+      "loss": 0.9542,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2635814946604828e-05,
+      "loss": 0.9341,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2632808969723927e-05,
+      "loss": 1.0382,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.262980273720485e-05,
+      "loss": 0.8727,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.26267962493395e-05,
+      "loss": 0.9966,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2623789506419792e-05,
+      "loss": 0.8632,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2620782508737678e-05,
+      "loss": 1.0038,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2617775256585126e-05,
+      "loss": 0.931,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2614767750254129e-05,
+      "loss": 0.9013,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2611759990036711e-05,
+      "loss": 1.0077,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2608751976224916e-05,
+      "loss": 0.984,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2605743709110816e-05,
+      "loss": 0.9886,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2602735188986498e-05,
+      "loss": 0.9848,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2599726416144087e-05,
+      "loss": 0.9542,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2596717390875721e-05,
+      "loss": 0.9864,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2593708113473573e-05,
+      "loss": 0.9363,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2590698584229834e-05,
+      "loss": 0.9644,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2587688803436719e-05,
+      "loss": 0.9542,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2584678771386467e-05,
+      "loss": 0.9565,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.258166848837135e-05,
+      "loss": 0.9844,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2578657954683651e-05,
+      "loss": 0.9097,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2575647170615687e-05,
+      "loss": 1.031,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2572636136459799e-05,
+      "loss": 0.9955,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2569624852508344e-05,
+      "loss": 0.9812,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2566613319053713e-05,
+      "loss": 1.0024,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2563601536388319e-05,
+      "loss": 0.9871,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2560589504804592e-05,
+      "loss": 0.9597,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2557577224594996e-05,
+      "loss": 0.979,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2554564696052011e-05,
+      "loss": 0.9298,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.255155191946815e-05,
+      "loss": 0.9662,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2548538895135942e-05,
+      "loss": 0.9368,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2545525623347943e-05,
+      "loss": 0.9609,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.254251210439673e-05,
+      "loss": 0.9948,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.253949833857491e-05,
+      "loss": 1.028,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2536484326175114e-05,
+      "loss": 1.0019,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.253347006748999e-05,
+      "loss": 1.0513,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2530455562812214e-05,
+      "loss": 1.0425,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2527440812434488e-05,
+      "loss": 0.998,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.252442581664953e-05,
+      "loss": 0.9636,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2521410575750094e-05,
+      "loss": 1.042,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2518395090028952e-05,
+      "loss": 0.992,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2515379359778894e-05,
+      "loss": 0.9957,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2512363385292739e-05,
+      "loss": 1.0024,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.250934716686333e-05,
+      "loss": 0.9346,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2506330704783533e-05,
+      "loss": 1.012,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2503313999346242e-05,
+      "loss": 1.0294,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2500297050844367e-05,
+      "loss": 0.9135,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2497279859570843e-05,
+      "loss": 0.906,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2494262425818637e-05,
+      "loss": 0.9536,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2491244749880722e-05,
+      "loss": 1.0009,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2488226832050116e-05,
+      "loss": 1.0659,
+      "step": 4546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2485208672619852e-05,
+      "loss": 0.909,
+      "step": 4547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2482190271882973e-05,
+      "loss": 1.043,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2479171630132563e-05,
+      "loss": 0.9921,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2476152747661727e-05,
+      "loss": 0.9699,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2473133624763583e-05,
+      "loss": 1.0459,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2470114261731288e-05,
+      "loss": 0.958,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2467094658858003e-05,
+      "loss": 0.9537,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.246407481643693e-05,
+      "loss": 0.9543,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2461054734761287e-05,
+      "loss": 0.961,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.245803441412431e-05,
+      "loss": 1.0105,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2455013854819266e-05,
+      "loss": 0.9569,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2451993057139445e-05,
+      "loss": 0.9719,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2448972021378153e-05,
+      "loss": 0.977,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2445950747828732e-05,
+      "loss": 0.9733,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2442929236784532e-05,
+      "loss": 1.0077,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2439907488538934e-05,
+      "loss": 0.9935,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2436885503385341e-05,
+      "loss": 0.9965,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.243386328161718e-05,
+      "loss": 0.9705,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2430840823527904e-05,
+      "loss": 1.0302,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2427818129410975e-05,
+      "loss": 0.956,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2424795199559898e-05,
+      "loss": 0.9511,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2421772034268187e-05,
+      "loss": 0.9526,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2418748633829379e-05,
+      "loss": 0.9604,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2415724998537042e-05,
+      "loss": 0.9942,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.241270112868476e-05,
+      "loss": 1.0313,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2409677024566145e-05,
+      "loss": 0.9471,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.240665268647483e-05,
+      "loss": 0.9907,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.240362811470446e-05,
+      "loss": 1.0619,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.240060330954872e-05,
+      "loss": 1.0071,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2397578271301312e-05,
+      "loss": 0.9369,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2394553000255951e-05,
+      "loss": 0.9714,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2391527496706389e-05,
+      "loss": 0.9325,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2388501760946391e-05,
+      "loss": 0.9961,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2385475793269744e-05,
+      "loss": 0.9379,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2382449593970268e-05,
+      "loss": 1.0071,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2379423163341791e-05,
+      "loss": 1.0248,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2376396501678173e-05,
+      "loss": 0.9391,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2373369609273299e-05,
+      "loss": 1.0535,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2370342486421064e-05,
+      "loss": 0.9644,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2367315133415396e-05,
+      "loss": 0.9841,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2364287550550244e-05,
+      "loss": 0.9489,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2361259738119575e-05,
+      "loss": 0.9677,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2358231696417387e-05,
+      "loss": 0.8615,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2355203425737683e-05,
+      "loss": 0.9099,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2352174926374509e-05,
+      "loss": 0.9086,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2349146198621917e-05,
+      "loss": 0.9905,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2346117242773993e-05,
+      "loss": 0.9843,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2343088059124839e-05,
+      "loss": 1.013,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2340058647968577e-05,
+      "loss": 1.0021,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2337029009599357e-05,
+      "loss": 0.9401,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2333999144311348e-05,
+      "loss": 0.9673,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2330969052398735e-05,
+      "loss": 0.9183,
+      "step": 4598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2327938734155743e-05,
+      "loss": 0.9488,
+      "step": 4599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2324908189876597e-05,
+      "loss": 0.9106,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2321877419855558e-05,
+      "loss": 0.9946,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2318846424386907e-05,
+      "loss": 0.972,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.231581520376494e-05,
+      "loss": 1.0334,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2312783758283981e-05,
+      "loss": 0.9973,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2309752088238382e-05,
+      "loss": 0.9263,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.23067201939225e-05,
+      "loss": 0.963,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2303688075630726e-05,
+      "loss": 0.9487,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2300655733657475e-05,
+      "loss": 0.9592,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.229762316829717e-05,
+      "loss": 0.971,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2294590379844268e-05,
+      "loss": 0.985,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2291557368593248e-05,
+      "loss": 0.9276,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2288524134838602e-05,
+      "loss": 1.0029,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2285490678874851e-05,
+      "loss": 0.9539,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2282457000996533e-05,
+      "loss": 0.9881,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.227942310149821e-05,
+      "loss": 0.9315,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2276388980674465e-05,
+      "loss": 0.9913,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2273354638819902e-05,
+      "loss": 0.9648,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.227032007622915e-05,
+      "loss": 0.9673,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2267285293196849e-05,
+      "loss": 0.9591,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2264250290017675e-05,
+      "loss": 0.965,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2261215066986311e-05,
+      "loss": 1.03,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2258179624397477e-05,
+      "loss": 0.9611,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2255143962545902e-05,
+      "loss": 1.0441,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2252108081726337e-05,
+      "loss": 0.9526,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.224907198223356e-05,
+      "loss": 1.069,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.224603566436237e-05,
+      "loss": 0.9414,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2242999128407578e-05,
+      "loss": 1.0094,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2239962374664029e-05,
+      "loss": 0.9573,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.223692540342658e-05,
+      "loss": 1.0309,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2233888214990113e-05,
+      "loss": 0.9745,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2230850809649533e-05,
+      "loss": 1.0103,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2227813187699757e-05,
+      "loss": 0.9749,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2224775349435733e-05,
+      "loss": 0.9606,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.222173729515243e-05,
+      "loss": 0.9762,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.221869902514483e-05,
+      "loss": 1.0393,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2215660539707936e-05,
+      "loss": 0.9535,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2212621839136787e-05,
+      "loss": 0.9755,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2209582923726424e-05,
+      "loss": 0.9675,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2206543793771919e-05,
+      "loss": 0.9684,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2203504449568361e-05,
+      "loss": 0.9027,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2200464891410866e-05,
+      "loss": 1.0103,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2197425119594563e-05,
+      "loss": 1.0491,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2194385134414608e-05,
+      "loss": 0.978,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.219134493616617e-05,
+      "loss": 0.908,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2188304525144448e-05,
+      "loss": 0.9454,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2185263901644653e-05,
+      "loss": 1.0177,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2182223065962026e-05,
+      "loss": 0.9437,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.217918201839182e-05,
+      "loss": 0.9313,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2176140759229312e-05,
+      "loss": 0.9908,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2173099288769799e-05,
+      "loss": 1.0206,
+      "step": 4650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2170057607308601e-05,
+      "loss": 0.9543,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2167015715141057e-05,
+      "loss": 0.8969,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2163973612562523e-05,
+      "loss": 0.9961,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.216093129986838e-05,
+      "loss": 1.017,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.215788877735403e-05,
+      "loss": 0.9673,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2154846045314885e-05,
+      "loss": 1.0668,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2151803104046396e-05,
+      "loss": 1.023,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.214875995384402e-05,
+      "loss": 1.0281,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2145716595003237e-05,
+      "loss": 0.9877,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.214267302781955e-05,
+      "loss": 0.959,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2139629252588477e-05,
+      "loss": 0.9012,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2136585269605558e-05,
+      "loss": 0.9265,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2133541079166366e-05,
+      "loss": 0.9751,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2130496681566475e-05,
+      "loss": 0.9861,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2127452077101486e-05,
+      "loss": 0.9817,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.212440726606703e-05,
+      "loss": 0.9768,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2121362248758738e-05,
+      "loss": 1.0152,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.211831702547228e-05,
+      "loss": 0.9818,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2115271596503336e-05,
+      "loss": 0.889,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2112225962147605e-05,
+      "loss": 0.9256,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.210918012270082e-05,
+      "loss": 0.9885,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.210613407845871e-05,
+      "loss": 1.0124,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2103087829717045e-05,
+      "loss": 0.9885,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2100041376771605e-05,
+      "loss": 0.9515,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2096994719918191e-05,
+      "loss": 0.9399,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.209394785945263e-05,
+      "loss": 1.0282,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2090900795670755e-05,
+      "loss": 0.9783,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2087853528868432e-05,
+      "loss": 1.0026,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2084806059341542e-05,
+      "loss": 0.9872,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2081758387385982e-05,
+      "loss": 1.0492,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.207871051329768e-05,
+      "loss": 1.0098,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2075662437372567e-05,
+      "loss": 0.8506,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2072614159906605e-05,
+      "loss": 0.9232,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2069565681195776e-05,
+      "loss": 1.0267,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2066517001536072e-05,
+      "loss": 0.9698,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.206346812122352e-05,
+      "loss": 0.9578,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2060419040554153e-05,
+      "loss": 0.9959,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2057369759824025e-05,
+      "loss": 0.8829,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2054320279329218e-05,
+      "loss": 0.9827,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2051270599365825e-05,
+      "loss": 0.9377,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2048220720229957e-05,
+      "loss": 0.9991,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2045170642217756e-05,
+      "loss": 0.8834,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.204212036562537e-05,
+      "loss": 0.9953,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2039069890748978e-05,
+      "loss": 0.837,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2036019217884768e-05,
+      "loss": 0.9934,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2032968347328952e-05,
+      "loss": 0.9067,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2029917279377761e-05,
+      "loss": 0.9512,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2026866014327446e-05,
+      "loss": 1.0092,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2023814552474273e-05,
+      "loss": 1.0229,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2020762894114535e-05,
+      "loss": 1.0251,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2017711039544535e-05,
+      "loss": 0.9924,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.20146589890606e-05,
+      "loss": 0.9912,
+      "step": 4702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2011606742959076e-05,
+      "loss": 0.981,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2008554301536328e-05,
+      "loss": 1.0424,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2005501665088742e-05,
+      "loss": 1.011,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2002448833912712e-05,
+      "loss": 0.9583,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1999395808304667e-05,
+      "loss": 0.959,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1996342588561042e-05,
+      "loss": 0.9753,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1993289174978298e-05,
+      "loss": 1.0505,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1990235567852917e-05,
+      "loss": 0.98,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1987181767481386e-05,
+      "loss": 0.9401,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1984127774160226e-05,
+      "loss": 0.9651,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1981073588185974e-05,
+      "loss": 0.988,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1978019209855174e-05,
+      "loss": 0.9293,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1974964639464401e-05,
+      "loss": 1.0439,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1971909877310253e-05,
+      "loss": 0.9316,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1968854923689328e-05,
+      "loss": 1.0598,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1965799778898258e-05,
+      "loss": 0.9963,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1962744443233686e-05,
+      "loss": 0.9485,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1959688916992279e-05,
+      "loss": 0.9539,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.195663320047072e-05,
+      "loss": 1.006,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1953577293965707e-05,
+      "loss": 1.0527,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1950521197773966e-05,
+      "loss": 0.9614,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1947464912192228e-05,
+      "loss": 0.9283,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1944408437517251e-05,
+      "loss": 0.9499,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1941351774045815e-05,
+      "loss": 0.9503,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1938294922074709e-05,
+      "loss": 0.9343,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1935237881900743e-05,
+      "loss": 0.9348,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1932180653820754e-05,
+      "loss": 0.9229,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1929123238131579e-05,
+      "loss": 0.9991,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1926065635130093e-05,
+      "loss": 0.9634,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1923007845113178e-05,
+      "loss": 0.952,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1919949868377735e-05,
+      "loss": 0.9754,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1916891705220689e-05,
+      "loss": 0.8353,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1913833355938973e-05,
+      "loss": 0.9312,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.191077482082955e-05,
+      "loss": 0.9368,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.190771610018939e-05,
+      "loss": 0.9803,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1904657194315486e-05,
+      "loss": 0.989,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1901598103504854e-05,
+      "loss": 1.0006,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1898538828054517e-05,
+      "loss": 0.9804,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1895479368261527e-05,
+      "loss": 0.925,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1892419724422946e-05,
+      "loss": 0.9808,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1889359896835857e-05,
+      "loss": 0.9614,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1886299885797357e-05,
+      "loss": 1.0013,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1883239691604572e-05,
+      "loss": 0.9395,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1880179314554629e-05,
+      "loss": 0.9623,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.187711875494469e-05,
+      "loss": 1.0119,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1874058013071923e-05,
+      "loss": 0.9855,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1870997089233515e-05,
+      "loss": 0.9122,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1867935983726676e-05,
+      "loss": 0.9717,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.186487469684863e-05,
+      "loss": 0.9543,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.186181322889662e-05,
+      "loss": 0.9672,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1858751580167904e-05,
+      "loss": 0.8825,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1855689750959759e-05,
+      "loss": 0.9751,
+      "step": 4754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1852627741569481e-05,
+      "loss": 0.9805,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1849565552294379e-05,
+      "loss": 1.0163,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1846503183431787e-05,
+      "loss": 0.9315,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1843440635279056e-05,
+      "loss": 0.9966,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.184037790813354e-05,
+      "loss": 0.9471,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1837315002292629e-05,
+      "loss": 0.9557,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1834251918053716e-05,
+      "loss": 1.029,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1831188655714225e-05,
+      "loss": 0.9462,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1828125215571587e-05,
+      "loss": 1.0232,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.182506159792325e-05,
+      "loss": 0.9886,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1821997803066687e-05,
+      "loss": 1.0285,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1818933831299381e-05,
+      "loss": 1.0289,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1815869682918835e-05,
+      "loss": 0.9575,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1812805358222571e-05,
+      "loss": 0.9325,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1809740857508123e-05,
+      "loss": 1.0527,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.180667618107305e-05,
+      "loss": 0.9727,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1803611329214917e-05,
+      "loss": 0.9236,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1800546302231317e-05,
+      "loss": 0.9737,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.179748110041985e-05,
+      "loss": 0.9814,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1794415724078147e-05,
+      "loss": 0.9159,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1791350173503839e-05,
+      "loss": 0.9498,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1788284448994588e-05,
+      "loss": 0.8751,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1785218550848065e-05,
+      "loss": 0.9382,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1782152479361956e-05,
+      "loss": 0.9817,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1779086234833971e-05,
+      "loss": 0.9594,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1776019817561834e-05,
+      "loss": 1.001,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1772953227843286e-05,
+      "loss": 1.0402,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1769886465976086e-05,
+      "loss": 1.007,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1766819532258e-05,
+      "loss": 0.9111,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1763752426986823e-05,
+      "loss": 0.9741,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1760685150460363e-05,
+      "loss": 1.0045,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1757617702976443e-05,
+      "loss": 1.0297,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1754550084832905e-05,
+      "loss": 1.0569,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.17514822963276e-05,
+      "loss": 0.9804,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1748414337758409e-05,
+      "loss": 0.9622,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1745346209423216e-05,
+      "loss": 0.8796,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1742277911619931e-05,
+      "loss": 0.9453,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1739209444646479e-05,
+      "loss": 0.8759,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1736140808800791e-05,
+      "loss": 0.9474,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1733072004380827e-05,
+      "loss": 0.9135,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1730003031684564e-05,
+      "loss": 0.9732,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1726933891009985e-05,
+      "loss": 1.1279,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1723864582655095e-05,
+      "loss": 0.8679,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1720795106917917e-05,
+      "loss": 1.0483,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1717725464096485e-05,
+      "loss": 1.0713,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.171465565448886e-05,
+      "loss": 0.9704,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1711585678393103e-05,
+      "loss": 0.962,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1708515536107299e-05,
+      "loss": 0.9938,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1705445227929562e-05,
+      "loss": 0.9649,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1702374754157998e-05,
+      "loss": 0.9695,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1699304115090743e-05,
+      "loss": 1.0193,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1696233311025957e-05,
+      "loss": 0.8993,
+      "step": 4806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1693162342261795e-05,
+      "loss": 0.9712,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1690091209096441e-05,
+      "loss": 1.0327,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1687019911828097e-05,
+      "loss": 0.9439,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1683948450754976e-05,
+      "loss": 1.0484,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.168087682617531e-05,
+      "loss": 1.0389,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1677805038387337e-05,
+      "loss": 0.9501,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1674733087689327e-05,
+      "loss": 1.0098,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1671660974379554e-05,
+      "loss": 0.9906,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.166858869875631e-05,
+      "loss": 1.0058,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1665516261117914e-05,
+      "loss": 0.8772,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1662443661762677e-05,
+      "loss": 1.0428,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1659370900988946e-05,
+      "loss": 0.9756,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.165629797909508e-05,
+      "loss": 0.9262,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.165322489637944e-05,
+      "loss": 0.9949,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.165015165314043e-05,
+      "loss": 0.9914,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.164707824967644e-05,
+      "loss": 0.9485,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1644004686285892e-05,
+      "loss": 0.9955,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1640930963267226e-05,
+      "loss": 1.0186,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.163785708091888e-05,
+      "loss": 0.9885,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1634783039539328e-05,
+      "loss": 1.0222,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.163170883942705e-05,
+      "loss": 1.0135,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.162863448088054e-05,
+      "loss": 1.0259,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1625559964198313e-05,
+      "loss": 0.9948,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1622485289678886e-05,
+      "loss": 0.9814,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1619410457620808e-05,
+      "loss": 1.0205,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1616335468322641e-05,
+      "loss": 1.0913,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1613260322082948e-05,
+      "loss": 0.9671,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1610185019200324e-05,
+      "loss": 1.0164,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.160710955997337e-05,
+      "loss": 0.9952,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1604033944700701e-05,
+      "loss": 0.9419,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1600958173680955e-05,
+      "loss": 0.9709,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1597882247212776e-05,
+      "loss": 1.0186,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1594806165594832e-05,
+      "loss": 0.9726,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.15917299291258e-05,
+      "loss": 1.0235,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1588653538104376e-05,
+      "loss": 0.9643,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1585576992829261e-05,
+      "loss": 0.9553,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1582500293599187e-05,
+      "loss": 1.0203,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1579423440712887e-05,
+      "loss": 0.9534,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1576346434469122e-05,
+      "loss": 1.0117,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1573269275166652e-05,
+      "loss": 0.9528,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1570191963104263e-05,
+      "loss": 1.0613,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1567114498580758e-05,
+      "loss": 0.9674,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.156403688189494e-05,
+      "loss": 0.986,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1560959113345649e-05,
+      "loss": 0.9185,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1557881193231717e-05,
+      "loss": 0.9417,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1554803121852005e-05,
+      "loss": 1.0449,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.155172489950539e-05,
+      "loss": 1.0323,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1548646526490749e-05,
+      "loss": 1.0392,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1545568003106985e-05,
+      "loss": 1.0085,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1542489329653024e-05,
+      "loss": 1.0132,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1539410506427786e-05,
+      "loss": 0.9304,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.153633153373022e-05,
+      "loss": 0.9484,
+      "step": 4858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1533252411859283e-05,
+      "loss": 0.9929,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1530173141113947e-05,
+      "loss": 1.0052,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.152709372179321e-05,
+      "loss": 1.0515,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1524014154196063e-05,
+      "loss": 1.0059,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1520934438621529e-05,
+      "loss": 1.0138,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1517854575368644e-05,
+      "loss": 0.8826,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1514774564736443e-05,
+      "loss": 0.9083,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1511694407023994e-05,
+      "loss": 0.9852,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1508614102530369e-05,
+      "loss": 1.0115,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1505533651554654e-05,
+      "loss": 0.991,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1502453054395961e-05,
+      "loss": 0.9715,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1499372311353398e-05,
+      "loss": 0.9786,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1496291422726098e-05,
+      "loss": 0.9468,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.149321038881321e-05,
+      "loss": 0.8682,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.149012920991389e-05,
+      "loss": 0.9399,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1487047886327314e-05,
+      "loss": 1.0322,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1483966418352669e-05,
+      "loss": 1.0008,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1480884806289151e-05,
+      "loss": 0.8856,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.147780305043599e-05,
+      "loss": 0.996,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1474721151092397e-05,
+      "loss": 1.0052,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.147163910855763e-05,
+      "loss": 0.9689,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1468556923130943e-05,
+      "loss": 0.9696,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1465474595111603e-05,
+      "loss": 1.0222,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.14623921247989e-05,
+      "loss": 1.0017,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.145930951249213e-05,
+      "loss": 1.0347,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1456226758490603e-05,
+      "loss": 0.8958,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1453143863093654e-05,
+      "loss": 0.9653,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1450060826600618e-05,
+      "loss": 0.9957,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.144697764931085e-05,
+      "loss": 0.9929,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1443894331523718e-05,
+      "loss": 0.9687,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.14408108735386e-05,
+      "loss": 0.9364,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1437727275654893e-05,
+      "loss": 1.0761,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1434643538172009e-05,
+      "loss": 0.983,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1431559661389362e-05,
+      "loss": 0.9816,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1428475645606399e-05,
+      "loss": 1.0302,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1425391491122557e-05,
+      "loss": 0.9387,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1422307198237306e-05,
+      "loss": 0.8672,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.141922276725012e-05,
+      "loss": 0.9755,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1416138198460488e-05,
+      "loss": 0.9827,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1413053492167915e-05,
+      "loss": 1.0486,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1409968648671914e-05,
+      "loss": 1.02,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1406883668272015e-05,
+      "loss": 0.9544,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1403798551267762e-05,
+      "loss": 0.8669,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.140071329795871e-05,
+      "loss": 0.9823,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1397627908644431e-05,
+      "loss": 0.9217,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.13945423836245e-05,
+      "loss": 1.0133,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1391456723198522e-05,
+      "loss": 1.002,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1388370927666102e-05,
+      "loss": 0.9909,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1385284997326855e-05,
+      "loss": 0.9387,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1382198932480429e-05,
+      "loss": 0.9343,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1379112733426461e-05,
+      "loss": 0.9698,
+      "step": 4909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1376026400464616e-05,
+      "loss": 0.961,
+      "step": 4910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1372939933894573e-05,
+      "loss": 0.9177,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.136985333401601e-05,
+      "loss": 0.95,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.136676660112863e-05,
+      "loss": 0.9349,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1363679735532151e-05,
+      "loss": 0.9131,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1360592737526292e-05,
+      "loss": 0.9722,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1357505607410797e-05,
+      "loss": 0.9926,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1354418345485413e-05,
+      "loss": 0.9309,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1351330952049908e-05,
+      "loss": 0.9675,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1348243427404055e-05,
+      "loss": 0.9463,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1345155771847646e-05,
+      "loss": 1.0052,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1342067985680485e-05,
+      "loss": 0.9555,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1338980069202388e-05,
+      "loss": 0.947,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1335892022713177e-05,
+      "loss": 0.8971,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1332803846512697e-05,
+      "loss": 1.0524,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1329715540900797e-05,
+      "loss": 0.9445,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1326627106177348e-05,
+      "loss": 0.926,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1323538542642227e-05,
+      "loss": 0.8956,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.132044985059532e-05,
+      "loss": 0.9649,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1317361030336535e-05,
+      "loss": 0.9216,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1314272082165785e-05,
+      "loss": 0.9126,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1311183006382998e-05,
+      "loss": 0.9217,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1308093803288119e-05,
+      "loss": 0.9831,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1305004473181094e-05,
+      "loss": 0.995,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.130191501636189e-05,
+      "loss": 1.0053,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1298825433130488e-05,
+      "loss": 0.9878,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1295735723786872e-05,
+      "loss": 0.925,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1292645888631051e-05,
+      "loss": 0.9707,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1289555927963032e-05,
+      "loss": 0.9933,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1286465842082843e-05,
+      "loss": 1.0361,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1283375631290528e-05,
+      "loss": 1.0444,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.128028529588613e-05,
+      "loss": 0.9736,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1277194836169714e-05,
+      "loss": 1.0096,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1274104252441358e-05,
+      "loss": 1.0352,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1271013545001144e-05,
+      "loss": 0.9833,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1267922714149176e-05,
+      "loss": 0.9794,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1264831760185562e-05,
+      "loss": 0.9262,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1261740683410424e-05,
+      "loss": 1.0181,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1258649484123895e-05,
+      "loss": 1.0073,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1255558162626128e-05,
+      "loss": 0.9222,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1252466719217274e-05,
+      "loss": 0.9816,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.124937515419751e-05,
+      "loss": 0.9661,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1246283467867012e-05,
+      "loss": 0.9462,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1243191660525979e-05,
+      "loss": 1.001,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1240099732474613e-05,
+      "loss": 1.0046,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.123700768401313e-05,
+      "loss": 0.971,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1233915515441765e-05,
+      "loss": 0.918,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1230823227060755e-05,
+      "loss": 1.0353,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1227730819170349e-05,
+      "loss": 0.9739,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.122463829207082e-05,
+      "loss": 1.0355,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1221545646062431e-05,
+      "loss": 1.0349,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.121845288144548e-05,
+      "loss": 0.9594,
+      "step": 4961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.121535999852026e-05,
+      "loss": 0.9403,
+      "step": 4962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1212266997587082e-05,
+      "loss": 0.9797,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1209173878946271e-05,
+      "loss": 1.0466,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1206080642898156e-05,
+      "loss": 0.9843,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1202987289743078e-05,
+      "loss": 1.0299,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1199893819781402e-05,
+      "loss": 1.0075,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1196800233313488e-05,
+      "loss": 0.9738,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1193706530639716e-05,
+      "loss": 0.8986,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1190612712060475e-05,
+      "loss": 0.9183,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1187518777876165e-05,
+      "loss": 0.9574,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1184424728387204e-05,
+      "loss": 0.9002,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1181330563894009e-05,
+      "loss": 1.0496,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1178236284697017e-05,
+      "loss": 0.9531,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1175141891096675e-05,
+      "loss": 0.9647,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1172047383393434e-05,
+      "loss": 0.9987,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1168952761887769e-05,
+      "loss": 0.9687,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1165858026880151e-05,
+      "loss": 0.998,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1162763178671078e-05,
+      "loss": 0.928,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1159668217561048e-05,
+      "loss": 1.0145,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.115657314385057e-05,
+      "loss": 1.0338,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.115347795784017e-05,
+      "loss": 1.0336,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1150382659830382e-05,
+      "loss": 0.9709,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1147287250121745e-05,
+      "loss": 0.9836,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1144191729014823e-05,
+      "loss": 0.913,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1141096096810174e-05,
+      "loss": 0.9357,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.113800035380838e-05,
+      "loss": 0.9537,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1134904500310029e-05,
+      "loss": 0.9972,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1131808536615712e-05,
+      "loss": 0.9809,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1128712463026048e-05,
+      "loss": 0.9858,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1125616279841654e-05,
+      "loss": 0.9736,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1122519987363156e-05,
+      "loss": 0.9617,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.11194235858912e-05,
+      "loss": 0.9661,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1116327075726436e-05,
+      "loss": 1.0762,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1113230457169526e-05,
+      "loss": 0.931,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1110133730521142e-05,
+      "loss": 0.9781,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1107036896081968e-05,
+      "loss": 1.0824,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.11039399541527e-05,
+      "loss": 1.0104,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1100842905034039e-05,
+      "loss": 0.9716,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.10977457490267e-05,
+      "loss": 0.9526,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1094648486431408e-05,
+      "loss": 0.9168,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.10915511175489e-05,
+      "loss": 0.9483,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.108845364267992e-05,
+      "loss": 1.0004,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1085356062125225e-05,
+      "loss": 0.9676,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.108225837618558e-05,
+      "loss": 0.9521,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1079160585161759e-05,
+      "loss": 0.9564,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1076062689354554e-05,
+      "loss": 0.9969,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.107296468906476e-05,
+      "loss": 0.9937,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1069866584593183e-05,
+      "loss": 0.8713,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.106676837624064e-05,
+      "loss": 0.9301,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1063670064307956e-05,
+      "loss": 1.0222,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1060571649095972e-05,
+      "loss": 1.0631,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1057473130905533e-05,
+      "loss": 0.9536,
+      "step": 5013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.10543745100375e-05,
+      "loss": 0.9468,
+      "step": 5014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1051275786792735e-05,
+      "loss": 0.9492,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1048176961472114e-05,
+      "loss": 1.0119,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1045078034376532e-05,
+      "loss": 0.9827,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1041979005806876e-05,
+      "loss": 0.9373,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1038879876064064e-05,
+      "loss": 0.9309,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1035780645449001e-05,
+      "loss": 0.9171,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1032681314262622e-05,
+      "loss": 1.0511,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1029581882805857e-05,
+      "loss": 0.8735,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1026482351379657e-05,
+      "loss": 0.934,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1023382720284973e-05,
+      "loss": 0.9952,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1020282989822774e-05,
+      "loss": 0.9257,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1017183160294033e-05,
+      "loss": 0.9734,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1014083231999738e-05,
+      "loss": 1.0386,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1010983205240878e-05,
+      "loss": 0.9392,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1007883080318456e-05,
+      "loss": 0.9107,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1004782857533488e-05,
+      "loss": 0.8981,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1001682537187e-05,
+      "loss": 0.9369,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.099858211958002e-05,
+      "loss": 0.9395,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.099548160501359e-05,
+      "loss": 0.9528,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0992380993788763e-05,
+      "loss": 0.949,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0989280286206595e-05,
+      "loss": 1.009,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0986179482568162e-05,
+      "loss": 0.9783,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0983078583174539e-05,
+      "loss": 0.9716,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0979977588326815e-05,
+      "loss": 0.9647,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0976876498326087e-05,
+      "loss": 0.9897,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0973775313473465e-05,
+      "loss": 0.8739,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.097067403407006e-05,
+      "loss": 1.0255,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0967572660417001e-05,
+      "loss": 1.0264,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0964471192815425e-05,
+      "loss": 0.9576,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0961369631566468e-05,
+      "loss": 0.981,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0958267976971289e-05,
+      "loss": 1.0402,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0955166229331048e-05,
+      "loss": 1.0118,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0952064388946912e-05,
+      "loss": 1.0557,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0948962456120068e-05,
+      "loss": 0.9455,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0945860431151702e-05,
+      "loss": 0.9883,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0942758314343007e-05,
+      "loss": 0.9029,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0939656105995198e-05,
+      "loss": 1.0162,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0936553806409482e-05,
+      "loss": 0.9934,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0933451415887086e-05,
+      "loss": 0.9599,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0930348934729249e-05,
+      "loss": 0.9866,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0927246363237205e-05,
+      "loss": 0.9713,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0924143701712211e-05,
+      "loss": 0.9793,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0921040950455521e-05,
+      "loss": 0.9333,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0917938109768404e-05,
+      "loss": 0.95,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0914835179952141e-05,
+      "loss": 0.8872,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0911732161308014e-05,
+      "loss": 1.0165,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0908629054137322e-05,
+      "loss": 0.9167,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0905525858741364e-05,
+      "loss": 0.9449,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0902422575421447e-05,
+      "loss": 0.8993,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.08993192044789e-05,
+      "loss": 0.9623,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0896215746215044e-05,
+      "loss": 0.952,
+      "step": 5065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.089311220093122e-05,
+      "loss": 1.0182,
+      "step": 5066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0890008568928772e-05,
+      "loss": 0.9695,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0886904850509052e-05,
+      "loss": 1.0062,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0883801045973425e-05,
+      "loss": 0.975,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0880697155623264e-05,
+      "loss": 0.955,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0877593179759937e-05,
+      "loss": 0.9853,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0874489118684846e-05,
+      "loss": 0.9172,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0871384972699372e-05,
+      "loss": 1.04,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.086828074210493e-05,
+      "loss": 1.0178,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0865176427202925e-05,
+      "loss": 1.0132,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0862072028294777e-05,
+      "loss": 0.8634,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0858967545681922e-05,
+      "loss": 0.969,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0855862979665788e-05,
+      "loss": 0.9285,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0852758330547822e-05,
+      "loss": 0.937,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0849653598629477e-05,
+      "loss": 0.9877,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0846548784212212e-05,
+      "loss": 0.8909,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0843443887597495e-05,
+      "loss": 1.0005,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0840338909086809e-05,
+      "loss": 0.989,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0837233848981632e-05,
+      "loss": 0.9084,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0834128707583455e-05,
+      "loss": 1.012,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0831023485193787e-05,
+      "loss": 0.9866,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0827918182114128e-05,
+      "loss": 0.9938,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0824812798645997e-05,
+      "loss": 0.9705,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0821707335090914e-05,
+      "loss": 0.9893,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0818601791750418e-05,
+      "loss": 0.9766,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0815496168926045e-05,
+      "loss": 0.98,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0812390466919337e-05,
+      "loss": 1.0244,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0809284686031855e-05,
+      "loss": 0.9357,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0806178826565162e-05,
+      "loss": 0.9537,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0803072888820823e-05,
+      "loss": 0.9991,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0799966873100419e-05,
+      "loss": 1.0687,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0796860779705533e-05,
+      "loss": 0.9793,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0793754608937758e-05,
+      "loss": 0.9944,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.07906483610987e-05,
+      "loss": 0.9393,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0787542036489955e-05,
+      "loss": 1.0319,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0784435635413152e-05,
+      "loss": 0.8574,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0781329158169902e-05,
+      "loss": 0.8976,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0778222605061841e-05,
+      "loss": 0.9881,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0775115976390607e-05,
+      "loss": 0.9521,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.077200927245784e-05,
+      "loss": 1.0366,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0768902493565197e-05,
+      "loss": 0.8967,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0765795640014332e-05,
+      "loss": 0.975,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0762688712106918e-05,
+      "loss": 1.0063,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0759581710144624e-05,
+      "loss": 0.9721,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0756474634429133e-05,
+      "loss": 0.8903,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0753367485262127e-05,
+      "loss": 0.9219,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0750260262945314e-05,
+      "loss": 0.915,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0747152967780385e-05,
+      "loss": 0.9985,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0744045600069055e-05,
+      "loss": 0.9896,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0740938160113042e-05,
+      "loss": 1.0318,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0737830648214063e-05,
+      "loss": 1.0348,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.073472306467385e-05,
+      "loss": 1.0374,
+      "step": 5117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0731615409794144e-05,
+      "loss": 0.9826,
+      "step": 5118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0728507683876688e-05,
+      "loss": 0.9416,
+      "step": 5119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0725399887223234e-05,
+      "loss": 0.877,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0722292020135536e-05,
+      "loss": 0.9749,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0719184082915364e-05,
+      "loss": 0.9593,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0716076075864486e-05,
+      "loss": 1.0142,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0712967999284682e-05,
+      "loss": 0.917,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.070985985347774e-05,
+      "loss": 0.8991,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0706751638745448e-05,
+      "loss": 0.9095,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0703643355389607e-05,
+      "loss": 0.9974,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0700535003712023e-05,
+      "loss": 1.0774,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.06974265840145e-05,
+      "loss": 0.9919,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.069431809659887e-05,
+      "loss": 0.9601,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0691209541766949e-05,
+      "loss": 0.985,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.068810091982057e-05,
+      "loss": 0.9626,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0684992231061576e-05,
+      "loss": 0.9342,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0681883475791803e-05,
+      "loss": 0.9346,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0678774654313112e-05,
+      "loss": 0.9322,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.067566576692735e-05,
+      "loss": 0.9742,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.067255681393639e-05,
+      "loss": 0.9771,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0669447795642103e-05,
+      "loss": 1.0193,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0666338712346357e-05,
+      "loss": 1.017,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.066322956435104e-05,
+      "loss": 0.933,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0660120351958044e-05,
+      "loss": 0.9912,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.065701107546926e-05,
+      "loss": 1.0099,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0653901735186591e-05,
+      "loss": 0.967,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.065079233141195e-05,
+      "loss": 0.9503,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0647682864447244e-05,
+      "loss": 0.914,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0644573334594395e-05,
+      "loss": 0.8964,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0641463742155332e-05,
+      "loss": 0.9582,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0638354087431986e-05,
+      "loss": 1.0415,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0635244370726297e-05,
+      "loss": 0.9442,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0632134592340204e-05,
+      "loss": 0.988,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0629024752575665e-05,
+      "loss": 0.9572,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0625914851734632e-05,
+      "loss": 0.9959,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0622804890119067e-05,
+      "loss": 0.863,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0619694868030943e-05,
+      "loss": 0.9866,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.061658478577223e-05,
+      "loss": 0.989,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0613474643644907e-05,
+      "loss": 0.9417,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0610364441950967e-05,
+      "loss": 0.95,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0607254180992391e-05,
+      "loss": 0.9845,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0604143861071183e-05,
+      "loss": 0.903,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0601033482489346e-05,
+      "loss": 0.9655,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0597923045548889e-05,
+      "loss": 1.027,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0594812550551826e-05,
+      "loss": 0.9894,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0591701997800172e-05,
+      "loss": 0.9784,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.058859138759596e-05,
+      "loss": 0.9283,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.058548072024122e-05,
+      "loss": 0.9766,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0582369996037985e-05,
+      "loss": 1.0174,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.05792592152883e-05,
+      "loss": 0.9937,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0576148378294213e-05,
+      "loss": 1.0073,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0573037485357776e-05,
+      "loss": 0.9371,
+      "step": 5169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.056992653678105e-05,
+      "loss": 0.9509,
+      "step": 5170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0566815532866096e-05,
+      "loss": 0.993,
+      "step": 5171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0563704473914986e-05,
+      "loss": 0.998,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0560593360229796e-05,
+      "loss": 0.9727,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0557482192112603e-05,
+      "loss": 0.9356,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0554370969865492e-05,
+      "loss": 0.9634,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0551259693790556e-05,
+      "loss": 0.9182,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0548148364189892e-05,
+      "loss": 0.9545,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0545036981365601e-05,
+      "loss": 0.975,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0541925545619786e-05,
+      "loss": 0.9125,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.053881405725456e-05,
+      "loss": 0.9163,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0535702516572039e-05,
+      "loss": 0.9247,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0532590923874349e-05,
+      "loss": 0.9217,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0529479279463612e-05,
+      "loss": 0.9722,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0526367583641958e-05,
+      "loss": 1.0129,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.052325583671153e-05,
+      "loss": 0.9385,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0520144038974468e-05,
+      "loss": 0.9599,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0517032190732913e-05,
+      "loss": 1.0404,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0513920292289021e-05,
+      "loss": 0.9768,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.051080834394495e-05,
+      "loss": 0.9277,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0507696346002857e-05,
+      "loss": 0.928,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0504584298764913e-05,
+      "loss": 1.0382,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0501472202533285e-05,
+      "loss": 1.0036,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0498360057610145e-05,
+      "loss": 0.8918,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0495247864297684e-05,
+      "loss": 0.9314,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0492135622898077e-05,
+      "loss": 0.9391,
+      "step": 5195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0489023333713522e-05,
+      "loss": 0.9244,
+      "step": 5196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0485910997046205e-05,
+      "loss": 0.8827,
+      "step": 5197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0482798613198328e-05,
+      "loss": 1.024,
+      "step": 5198
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0479686182472097e-05,
+      "loss": 0.858,
+      "step": 5199
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.047657370516972e-05,
+      "loss": 0.9142,
+      "step": 5200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0473461181593405e-05,
+      "loss": 0.941,
+      "step": 5201
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0470348612045376e-05,
+      "loss": 0.8168,
+      "step": 5202
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0467235996827847e-05,
+      "loss": 0.8615,
+      "step": 5203
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0464123336243049e-05,
+      "loss": 0.9601,
+      "step": 5204
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.046101063059321e-05,
+      "loss": 0.9722,
+      "step": 5205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0457897880180566e-05,
+      "loss": 0.781,
+      "step": 5206
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0454785085307357e-05,
+      "loss": 0.9202,
+      "step": 5207
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0451672246275826e-05,
+      "loss": 0.8949,
+      "step": 5208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0448559363388218e-05,
+      "loss": 0.9095,
+      "step": 5209
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0445446436946788e-05,
+      "loss": 0.9227,
+      "step": 5210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.044233346725379e-05,
+      "loss": 0.8987,
+      "step": 5211
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0439220454611486e-05,
+      "loss": 0.8477,
+      "step": 5212
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.043610739932214e-05,
+      "loss": 0.9668,
+      "step": 5213
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0432994301688021e-05,
+      "loss": 0.8134,
+      "step": 5214
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0429881162011404e-05,
+      "loss": 0.8163,
+      "step": 5215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0426767980594559e-05,
+      "loss": 0.8603,
+      "step": 5216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0423654757739776e-05,
+      "loss": 0.8717,
+      "step": 5217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0420541493749332e-05,
+      "loss": 0.8503,
+      "step": 5218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.041742818892552e-05,
+      "loss": 0.8781,
+      "step": 5219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0414314843570634e-05,
+      "loss": 0.8515,
+      "step": 5220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.041120145798697e-05,
+      "loss": 0.8414,
+      "step": 5221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0408088032476822e-05,
+      "loss": 0.8643,
+      "step": 5222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0404974567342507e-05,
+      "loss": 0.8862,
+      "step": 5223
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0401861062886324e-05,
+      "loss": 0.7693,
+      "step": 5224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0398747519410593e-05,
+      "loss": 0.9033,
+      "step": 5225
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0395633937217622e-05,
+      "loss": 0.8305,
+      "step": 5226
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0392520316609735e-05,
+      "loss": 0.8567,
+      "step": 5227
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0389406657889254e-05,
+      "loss": 0.8572,
+      "step": 5228
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0386292961358508e-05,
+      "loss": 0.7887,
+      "step": 5229
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0383179227319826e-05,
+      "loss": 0.7954,
+      "step": 5230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0380065456075548e-05,
+      "loss": 0.8663,
+      "step": 5231
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0376951647928007e-05,
+      "loss": 0.8755,
+      "step": 5232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0373837803179543e-05,
+      "loss": 0.8779,
+      "step": 5233
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0370723922132506e-05,
+      "loss": 0.9097,
+      "step": 5234
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0367610005089244e-05,
+      "loss": 0.9144,
+      "step": 5235
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.036449605235211e-05,
+      "loss": 0.8897,
+      "step": 5236
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0361382064223456e-05,
+      "loss": 0.8116,
+      "step": 5237
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0358268041005644e-05,
+      "loss": 0.9289,
+      "step": 5238
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0355153983001038e-05,
+      "loss": 0.8441,
+      "step": 5239
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0352039890511997e-05,
+      "loss": 0.9119,
+      "step": 5240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0348925763840902e-05,
+      "loss": 0.8015,
+      "step": 5241
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.034581160329012e-05,
+      "loss": 0.9174,
+      "step": 5242
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0342697409162023e-05,
+      "loss": 0.7957,
+      "step": 5243
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0339583181758997e-05,
+      "loss": 0.9899,
+      "step": 5244
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.033646892138342e-05,
+      "loss": 0.824,
+      "step": 5245
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.033335462833768e-05,
+      "loss": 0.8822,
+      "step": 5246
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0330240302924163e-05,
+      "loss": 0.8145,
+      "step": 5247
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0327125945445265e-05,
+      "loss": 0.8399,
+      "step": 5248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.032401155620338e-05,
+      "loss": 0.8506,
+      "step": 5249
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0320897135500904e-05,
+      "loss": 0.843,
+      "step": 5250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0317782683640238e-05,
+      "loss": 0.948,
+      "step": 5251
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0314668200923791e-05,
+      "loss": 0.9031,
+      "step": 5252
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0311553687653969e-05,
+      "loss": 0.8949,
+      "step": 5253
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0308439144133177e-05,
+      "loss": 0.8842,
+      "step": 5254
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0305324570663835e-05,
+      "loss": 0.8028,
+      "step": 5255
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0302209967548354e-05,
+      "loss": 0.8284,
+      "step": 5256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0299095335089155e-05,
+      "loss": 0.8856,
+      "step": 5257
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.029598067358866e-05,
+      "loss": 0.9,
+      "step": 5258
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0292865983349292e-05,
+      "loss": 1.0109,
+      "step": 5259
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0289751264673485e-05,
+      "loss": 0.7798,
+      "step": 5260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.028663651786366e-05,
+      "loss": 0.8763,
+      "step": 5261
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0283521743222256e-05,
+      "loss": 0.9147,
+      "step": 5262
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0280406941051704e-05,
+      "loss": 0.8358,
+      "step": 5263
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0277292111654447e-05,
+      "loss": 0.9044,
+      "step": 5264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0274177255332926e-05,
+      "loss": 0.8186,
+      "step": 5265
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0271062372389582e-05,
+      "loss": 0.8025,
+      "step": 5266
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0267947463126859e-05,
+      "loss": 0.8381,
+      "step": 5267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0264832527847212e-05,
+      "loss": 0.8525,
+      "step": 5268
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0261717566853087e-05,
+      "loss": 0.8655,
+      "step": 5269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0258602580446941e-05,
+      "loss": 0.8211,
+      "step": 5270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0255487568931226e-05,
+      "loss": 0.9321,
+      "step": 5271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0252372532608405e-05,
+      "loss": 0.8757,
+      "step": 5272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.024925747178094e-05,
+      "loss": 0.925,
+      "step": 5273
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.024614238675129e-05,
+      "loss": 0.8133,
+      "step": 5274
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0243027277821922e-05,
+      "loss": 0.9296,
+      "step": 5275
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0239912145295303e-05,
+      "loss": 0.9175,
+      "step": 5276
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0236796989473906e-05,
+      "loss": 0.8473,
+      "step": 5277
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0233681810660207e-05,
+      "loss": 0.8438,
+      "step": 5278
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0230566609156675e-05,
+      "loss": 0.8025,
+      "step": 5279
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0227451385265788e-05,
+      "loss": 0.8973,
+      "step": 5280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0224336139290025e-05,
+      "loss": 0.7702,
+      "step": 5281
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.022122087153187e-05,
+      "loss": 0.8492,
+      "step": 5282
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0218105582293804e-05,
+      "loss": 0.906,
+      "step": 5283
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0214990271878319e-05,
+      "loss": 0.8809,
+      "step": 5284
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.021187494058789e-05,
+      "loss": 0.8824,
+      "step": 5285
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0208759588725016e-05,
+      "loss": 0.8802,
+      "step": 5286
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.020564421659219e-05,
+      "loss": 0.8625,
+      "step": 5287
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0202528824491899e-05,
+      "loss": 0.8409,
+      "step": 5288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0199413412726646e-05,
+      "loss": 0.8433,
+      "step": 5289
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0196297981598921e-05,
+      "loss": 0.843,
+      "step": 5290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.019318253141123e-05,
+      "loss": 0.9069,
+      "step": 5291
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.019006706246607e-05,
+      "loss": 0.8729,
+      "step": 5292
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0186951575065942e-05,
+      "loss": 0.9054,
+      "step": 5293
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.018383606951336e-05,
+      "loss": 0.8249,
+      "step": 5294
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.018072054611082e-05,
+      "loss": 0.922,
+      "step": 5295
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0177605005160837e-05,
+      "loss": 0.9543,
+      "step": 5296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0174489446965922e-05,
+      "loss": 0.7426,
+      "step": 5297
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0171373871828578e-05,
+      "loss": 0.8989,
+      "step": 5298
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0168258280051331e-05,
+      "loss": 0.8995,
+      "step": 5299
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0165142671936685e-05,
+      "loss": 0.9519,
+      "step": 5300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0162027047787162e-05,
+      "loss": 0.8865,
+      "step": 5301
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0158911407905279e-05,
+      "loss": 0.8632,
+      "step": 5302
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0155795752593556e-05,
+      "loss": 0.8909,
+      "step": 5303
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0152680082154514e-05,
+      "loss": 0.8109,
+      "step": 5304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0149564396890674e-05,
+      "loss": 0.8928,
+      "step": 5305
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0146448697104561e-05,
+      "loss": 0.8887,
+      "step": 5306
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0143332983098703e-05,
+      "loss": 0.8811,
+      "step": 5307
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0140217255175626e-05,
+      "loss": 0.8565,
+      "step": 5308
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0137101513637855e-05,
+      "loss": 0.8752,
+      "step": 5309
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.013398575878792e-05,
+      "loss": 0.8484,
+      "step": 5310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0130869990928356e-05,
+      "loss": 0.9324,
+      "step": 5311
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0127754210361694e-05,
+      "loss": 0.8593,
+      "step": 5312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0124638417390464e-05,
+      "loss": 0.9034,
+      "step": 5313
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0121522612317204e-05,
+      "loss": 0.8993,
+      "step": 5314
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0118406795444447e-05,
+      "loss": 0.9182,
+      "step": 5315
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.011529096707473e-05,
+      "loss": 0.7334,
+      "step": 5316
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0112175127510591e-05,
+      "loss": 0.787,
+      "step": 5317
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0109059277054574e-05,
+      "loss": 0.8799,
+      "step": 5318
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0105943416009212e-05,
+      "loss": 0.9235,
+      "step": 5319
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.010282754467705e-05,
+      "loss": 0.9037,
+      "step": 5320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0099711663360628e-05,
+      "loss": 0.9061,
+      "step": 5321
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0096595772362492e-05,
+      "loss": 0.8751,
+      "step": 5322
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0093479871985186e-05,
+      "loss": 0.8605,
+      "step": 5323
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0090363962531251e-05,
+      "loss": 0.8242,
+      "step": 5324
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0087248044303237e-05,
+      "loss": 0.8856,
+      "step": 5325
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0084132117603689e-05,
+      "loss": 0.9424,
+      "step": 5326
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0081016182735152e-05,
+      "loss": 0.8774,
+      "step": 5327
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0077900240000181e-05,
+      "loss": 0.8887,
+      "step": 5328
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0074784289701318e-05,
+      "loss": 0.8874,
+      "step": 5329
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0071668332141115e-05,
+      "loss": 0.8701,
+      "step": 5330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0068552367622127e-05,
+      "loss": 0.879,
+      "step": 5331
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0065436396446899e-05,
+      "loss": 0.8518,
+      "step": 5332
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0062320418917986e-05,
+      "loss": 0.9143,
+      "step": 5333
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0059204435337938e-05,
+      "loss": 0.7975,
+      "step": 5334
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0056088446009309e-05,
+      "loss": 0.8621,
+      "step": 5335
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0052972451234656e-05,
+      "loss": 0.8463,
+      "step": 5336
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.004985645131653e-05,
+      "loss": 0.9042,
+      "step": 5337
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0046740446557485e-05,
+      "loss": 0.902,
+      "step": 5338
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0043624437260078e-05,
+      "loss": 0.8949,
+      "step": 5339
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0040508423726865e-05,
+      "loss": 0.8725,
+      "step": 5340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0037392406260398e-05,
+      "loss": 0.826,
+      "step": 5341
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0034276385163238e-05,
+      "loss": 0.8561,
+      "step": 5342
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.003116036073794e-05,
+      "loss": 0.8304,
+      "step": 5343
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0028044333287056e-05,
+      "loss": 0.8698,
+      "step": 5344
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0024928303113152e-05,
+      "loss": 0.845,
+      "step": 5345
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.002181227051878e-05,
+      "loss": 0.8371,
+      "step": 5346
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0018696235806504e-05,
+      "loss": 0.9178,
+      "step": 5347
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0015580199278873e-05,
+      "loss": 0.8544,
+      "step": 5348
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0012464161238448e-05,
+      "loss": 0.9003,
+      "step": 5349
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0009348121987795e-05,
+      "loss": 0.864,
+      "step": 5350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.000623208182946e-05,
+      "loss": 0.9245,
+      "step": 5351
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.000311604106601e-05,
+      "loss": 0.8857,
+      "step": 5352
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.8203,
+      "step": 5353
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.996883958933993e-06,
+      "loss": 0.8819,
+      "step": 5354
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.993767918170544e-06,
+      "loss": 0.9418,
+      "step": 5355
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99065187801221e-06,
+      "loss": 0.8674,
+      "step": 5356
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.987535838761553e-06,
+      "loss": 0.8415,
+      "step": 5357
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.984419800721132e-06,
+      "loss": 0.8262,
+      "step": 5358
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.981303764193498e-06,
+      "loss": 0.8761,
+      "step": 5359
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.978187729481218e-06,
+      "loss": 0.77,
+      "step": 5360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.975071696886848e-06,
+      "loss": 0.8099,
+      "step": 5361
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.971955666712945e-06,
+      "loss": 0.8378,
+      "step": 5362
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.968839639262064e-06,
+      "loss": 0.8463,
+      "step": 5363
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.965723614836764e-06,
+      "loss": 0.8914,
+      "step": 5364
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.962607593739605e-06,
+      "loss": 0.8572,
+      "step": 5365
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.959491576273139e-06,
+      "loss": 0.8971,
+      "step": 5366
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.956375562739923e-06,
+      "loss": 0.8768,
+      "step": 5367
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.95325955344252e-06,
+      "loss": 0.9075,
+      "step": 5368
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.950143548683475e-06,
+      "loss": 0.8983,
+      "step": 5369
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.947027548765347e-06,
+      "loss": 0.7842,
+      "step": 5370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.943911553990691e-06,
+      "loss": 0.8679,
+      "step": 5371
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.940795564662064e-06,
+      "loss": 0.8326,
+      "step": 5372
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.937679581082017e-06,
+      "loss": 0.8824,
+      "step": 5373
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.934563603553103e-06,
+      "loss": 0.8979,
+      "step": 5374
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.931447632377874e-06,
+      "loss": 0.8794,
+      "step": 5375
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.928331667858886e-06,
+      "loss": 0.8978,
+      "step": 5376
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.925215710298684e-06,
+      "loss": 0.9503,
+      "step": 5377
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.922099759999822e-06,
+      "loss": 0.8999,
+      "step": 5378
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.918983817264851e-06,
+      "loss": 0.8559,
+      "step": 5379
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.915867882396314e-06,
+      "loss": 0.8713,
+      "step": 5380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.912751955696766e-06,
+      "loss": 0.8828,
+      "step": 5381
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.909636037468754e-06,
+      "loss": 0.8178,
+      "step": 5382
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.906520128014817e-06,
+      "loss": 0.8771,
+      "step": 5383
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.90340422763751e-06,
+      "loss": 0.8058,
+      "step": 5384
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.900288336639372e-06,
+      "loss": 0.8983,
+      "step": 5385
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.897172455322953e-06,
+      "loss": 0.8816,
+      "step": 5386
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.894056583990791e-06,
+      "loss": 0.9063,
+      "step": 5387
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.890940722945429e-06,
+      "loss": 0.8858,
+      "step": 5388
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.887824872489412e-06,
+      "loss": 0.9171,
+      "step": 5389
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.884709032925274e-06,
+      "loss": 0.8915,
+      "step": 5390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.881593204555558e-06,
+      "loss": 0.8716,
+      "step": 5391
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.878477387682801e-06,
+      "loss": 0.9403,
+      "step": 5392
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.875361582609541e-06,
+      "loss": 0.929,
+      "step": 5393
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.872245789638308e-06,
+      "loss": 0.8702,
+      "step": 5394
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.869130009071644e-06,
+      "loss": 0.8993,
+      "step": 5395
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.866014241212078e-06,
+      "loss": 0.9006,
+      "step": 5396
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.862898486362149e-06,
+      "loss": 0.9674,
+      "step": 5397
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.859782744824376e-06,
+      "loss": 0.8989,
+      "step": 5398
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.856667016901299e-06,
+      "loss": 0.858,
+      "step": 5399
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.85355130289544e-06,
+      "loss": 0.9254,
+      "step": 5400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.85043560310933e-06,
+      "loss": 0.9213,
+      "step": 5401
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.84731991784549e-06,
+      "loss": 0.8463,
+      "step": 5402
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.844204247406449e-06,
+      "loss": 0.8501,
+      "step": 5403
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.841088592094726e-06,
+      "loss": 0.9327,
+      "step": 5404
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.837972952212841e-06,
+      "loss": 0.9369,
+      "step": 5405
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.834857328063316e-06,
+      "loss": 0.8771,
+      "step": 5406
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.831741719948672e-06,
+      "loss": 0.8593,
+      "step": 5407
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.828626128171422e-06,
+      "loss": 0.925,
+      "step": 5408
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.825510553034081e-06,
+      "loss": 0.8928,
+      "step": 5409
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.822394994839164e-06,
+      "loss": 0.7553,
+      "step": 5410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.819279453889181e-06,
+      "loss": 0.8233,
+      "step": 5411
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.816163930486643e-06,
+      "loss": 0.9207,
+      "step": 5412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.81304842493406e-06,
+      "loss": 0.8506,
+      "step": 5413
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.809932937533935e-06,
+      "loss": 0.8766,
+      "step": 5414
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.806817468588774e-06,
+      "loss": 0.8799,
+      "step": 5415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.803702018401084e-06,
+      "loss": 0.889,
+      "step": 5416
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.800586587273359e-06,
+      "loss": 0.8501,
+      "step": 5417
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.797471175508101e-06,
+      "loss": 0.9237,
+      "step": 5418
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.79435578340781e-06,
+      "loss": 0.9083,
+      "step": 5419
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791240411274982e-06,
+      "loss": 0.923,
+      "step": 5420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.788125059412113e-06,
+      "loss": 0.8762,
+      "step": 5421
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.785009728121686e-06,
+      "loss": 0.845,
+      "step": 5422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.781894417706197e-06,
+      "loss": 0.9344,
+      "step": 5423
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.778779128468133e-06,
+      "loss": 0.9585,
+      "step": 5424
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.775663860709977e-06,
+      "loss": 0.8905,
+      "step": 5425
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.772548614734217e-06,
+      "loss": 0.8839,
+      "step": 5426
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.76943339084333e-06,
+      "loss": 0.8962,
+      "step": 5427
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.766318189339798e-06,
+      "loss": 0.8689,
+      "step": 5428
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.763203010526094e-06,
+      "loss": 0.9553,
+      "step": 5429
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.760087854704697e-06,
+      "loss": 0.8577,
+      "step": 5430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.756972722178082e-06,
+      "loss": 0.8711,
+      "step": 5431
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.753857613248714e-06,
+      "loss": 0.8629,
+      "step": 5432
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.750742528219063e-06,
+      "loss": 0.9312,
+      "step": 5433
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.747627467391596e-06,
+      "loss": 0.869,
+      "step": 5434
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.744512431068776e-06,
+      "loss": 0.8902,
+      "step": 5435
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.741397419553062e-06,
+      "loss": 0.8335,
+      "step": 5436
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.738282433146917e-06,
+      "loss": 0.8491,
+      "step": 5437
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.735167472152793e-06,
+      "loss": 0.8614,
+      "step": 5438
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.732052536873143e-06,
+      "loss": 0.9562,
+      "step": 5439
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.728937627610425e-06,
+      "loss": 0.9232,
+      "step": 5440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.725822744667075e-06,
+      "loss": 0.9539,
+      "step": 5441
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.722707888345553e-06,
+      "loss": 0.9021,
+      "step": 5442
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.719593058948296e-06,
+      "loss": 0.8705,
+      "step": 5443
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.716478256777749e-06,
+      "loss": 0.9666,
+      "step": 5444
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.713363482136344e-06,
+      "loss": 0.8545,
+      "step": 5445
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.710248735326519e-06,
+      "loss": 0.8718,
+      "step": 5446
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.70713401665071e-06,
+      "loss": 0.9632,
+      "step": 5447
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.704019326411344e-06,
+      "loss": 0.9664,
+      "step": 5448
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.700904664910848e-06,
+      "loss": 0.8669,
+      "step": 5449
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.697790032451651e-06,
+      "loss": 0.8813,
+      "step": 5450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.694675429336172e-06,
+      "loss": 0.8319,
+      "step": 5451
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.691560855866826e-06,
+      "loss": 0.8851,
+      "step": 5452
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.688446312346033e-06,
+      "loss": 0.8372,
+      "step": 5453
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.685331799076208e-06,
+      "loss": 0.8319,
+      "step": 5454
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.682217316359764e-06,
+      "loss": 0.9031,
+      "step": 5455
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.6791028644991e-06,
+      "loss": 0.8572,
+      "step": 5456
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.675988443796622e-06,
+      "loss": 0.9198,
+      "step": 5457
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.672874054554738e-06,
+      "loss": 0.9231,
+      "step": 5458
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.66975969707584e-06,
+      "loss": 0.9283,
+      "step": 5459
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.666645371662324e-06,
+      "loss": 0.95,
+      "step": 5460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.663531078616585e-06,
+      "loss": 0.8235,
+      "step": 5461
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.660416818241007e-06,
+      "loss": 0.9452,
+      "step": 5462
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.65730259083798e-06,
+      "loss": 0.8975,
+      "step": 5463
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.654188396709882e-06,
+      "loss": 0.8989,
+      "step": 5464
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.6510742361591e-06,
+      "loss": 0.8265,
+      "step": 5465
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.647960109488003e-06,
+      "loss": 0.9034,
+      "step": 5466
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.644846016998966e-06,
+      "loss": 0.8996,
+      "step": 5467
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.64173195899436e-06,
+      "loss": 0.9121,
+      "step": 5468
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.638617935776548e-06,
+      "loss": 0.902,
+      "step": 5469
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.635503947647894e-06,
+      "loss": 0.8964,
+      "step": 5470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.63238999491076e-06,
+      "loss": 0.9045,
+      "step": 5471
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.629276077867497e-06,
+      "loss": 0.8488,
+      "step": 5472
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.62616219682046e-06,
+      "loss": 0.8525,
+      "step": 5473
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.623048352071998e-06,
+      "loss": 0.8701,
+      "step": 5474
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.619934543924456e-06,
+      "loss": 0.9319,
+      "step": 5475
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.616820772680174e-06,
+      "loss": 0.8507,
+      "step": 5476
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.613707038641494e-06,
+      "loss": 0.9312,
+      "step": 5477
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.610593342110746e-06,
+      "loss": 0.8702,
+      "step": 5478
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.607479683390268e-06,
+      "loss": 0.9326,
+      "step": 5479
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.604366062782381e-06,
+      "loss": 0.9034,
+      "step": 5480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.601252480589409e-06,
+      "loss": 0.8663,
+      "step": 5481
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.598138937113677e-06,
+      "loss": 0.8727,
+      "step": 5482
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.595025432657496e-06,
+      "loss": 0.8588,
+      "step": 5483
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.59191196752318e-06,
+      "loss": 0.8131,
+      "step": 5484
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.588798542013035e-06,
+      "loss": 0.8228,
+      "step": 5485
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.58568515642937e-06,
+      "loss": 0.8708,
+      "step": 5486
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.582571811074485e-06,
+      "loss": 0.91,
+      "step": 5487
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.579458506250668e-06,
+      "loss": 0.8227,
+      "step": 5488
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.576345242260227e-06,
+      "loss": 0.9346,
+      "step": 5489
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.573232019405441e-06,
+      "loss": 0.8882,
+      "step": 5490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.5701188379886e-06,
+      "loss": 0.9204,
+      "step": 5491
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.567005698311982e-06,
+      "loss": 0.846,
+      "step": 5492
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.563892600677861e-06,
+      "loss": 0.8615,
+      "step": 5493
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.560779545388517e-06,
+      "loss": 0.9003,
+      "step": 5494
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.557666532746214e-06,
+      "loss": 0.9271,
+      "step": 5495
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.554553563053217e-06,
+      "loss": 0.935,
+      "step": 5496
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.551440636611785e-06,
+      "loss": 0.853,
+      "step": 5497
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.548327753724181e-06,
+      "loss": 0.8836,
+      "step": 5498
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.545214914692643e-06,
+      "loss": 0.8547,
+      "step": 5499
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.542102119819436e-06,
+      "loss": 0.9579,
+      "step": 5500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.538989369406792e-06,
+      "loss": 0.85,
+      "step": 5501
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.535876663756955e-06,
+      "loss": 0.7715,
+      "step": 5502
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.532764003172156e-06,
+      "loss": 0.8492,
+      "step": 5503
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.529651387954628e-06,
+      "loss": 0.9771,
+      "step": 5504
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.526538818406598e-06,
+      "loss": 0.7609,
+      "step": 5505
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.523426294830284e-06,
+      "loss": 0.8752,
+      "step": 5506
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.520313817527905e-06,
+      "loss": 0.8504,
+      "step": 5507
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.517201386801675e-06,
+      "loss": 0.9088,
+      "step": 5508
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.5140890029538e-06,
+      "loss": 0.9652,
+      "step": 5509
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.510976666286484e-06,
+      "loss": 0.8605,
+      "step": 5510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.507864377101923e-06,
+      "loss": 0.8652,
+      "step": 5511
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.504752135702318e-06,
+      "loss": 0.9377,
+      "step": 5512
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.501639942389855e-06,
+      "loss": 0.904,
+      "step": 5513
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.498527797466718e-06,
+      "loss": 0.8446,
+      "step": 5514
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.49541570123509e-06,
+      "loss": 0.8696,
+      "step": 5515
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.492303653997146e-06,
+      "loss": 0.8434,
+      "step": 5516
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.489191656055054e-06,
+      "loss": 0.9179,
+      "step": 5517
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.48607970771098e-06,
+      "loss": 0.853,
+      "step": 5518
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.482967809267092e-06,
+      "loss": 0.8307,
+      "step": 5519
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.479855961025538e-06,
+      "loss": 0.7846,
+      "step": 5520
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.476744163288473e-06,
+      "loss": 0.951,
+      "step": 5521
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.473632416358045e-06,
+      "loss": 0.8723,
+      "step": 5522
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.470520720536391e-06,
+      "loss": 0.7881,
+      "step": 5523
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.467409076125653e-06,
+      "loss": 0.8509,
+      "step": 5524
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.464297483427961e-06,
+      "loss": 0.9641,
+      "step": 5525
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.461185942745443e-06,
+      "loss": 0.8111,
+      "step": 5526
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.458074454380217e-06,
+      "loss": 0.862,
+      "step": 5527
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.454963018634402e-06,
+      "loss": 0.9034,
+      "step": 5528
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.451851635810111e-06,
+      "loss": 0.9532,
+      "step": 5529
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.448740306209447e-06,
+      "loss": 0.9707,
+      "step": 5530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.44562903013451e-06,
+      "loss": 0.8334,
+      "step": 5531
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.442517807887402e-06,
+      "loss": 0.8159,
+      "step": 5532
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.439406639770209e-06,
+      "loss": 0.8386,
+      "step": 5533
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.436295526085016e-06,
+      "loss": 0.8908,
+      "step": 5534
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.433184467133904e-06,
+      "loss": 0.8171,
+      "step": 5535
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.430073463218952e-06,
+      "loss": 0.9128,
+      "step": 5536
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.426962514642225e-06,
+      "loss": 0.8897,
+      "step": 5537
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.423851621705789e-06,
+      "loss": 0.8277,
+      "step": 5538
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.4207407847117e-06,
+      "loss": 0.8502,
+      "step": 5539
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.41763000396202e-06,
+      "loss": 0.8592,
+      "step": 5540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.414519279758785e-06,
+      "loss": 0.8076,
+      "step": 5541
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.411408612404043e-06,
+      "loss": 0.906,
+      "step": 5542
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.408298002199832e-06,
+      "loss": 0.8073,
+      "step": 5543
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.40518744944818e-06,
+      "loss": 0.8658,
+      "step": 5544
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.402076954451116e-06,
+      "loss": 0.8531,
+      "step": 5545
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.398966517510654e-06,
+      "loss": 0.9205,
+      "step": 5546
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.395856138928818e-06,
+      "loss": 0.8222,
+      "step": 5547
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.39274581900761e-06,
+      "loss": 0.8535,
+      "step": 5548
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.389635558049038e-06,
+      "loss": 0.8936,
+      "step": 5549
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.386525356355095e-06,
+      "loss": 0.861,
+      "step": 5550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.383415214227773e-06,
+      "loss": 0.7989,
+      "step": 5551
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.380305131969059e-06,
+      "loss": 0.854,
+      "step": 5552
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.377195109880936e-06,
+      "loss": 0.8268,
+      "step": 5553
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.374085148265372e-06,
+      "loss": 0.8464,
+      "step": 5554
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.370975247424338e-06,
+      "loss": 0.8459,
+      "step": 5555
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.3678654076598e-06,
+      "loss": 0.888,
+      "step": 5556
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.36475562927371e-06,
+      "loss": 0.8661,
+      "step": 5557
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.361645912568015e-06,
+      "loss": 0.8262,
+      "step": 5558
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.35853625784467e-06,
+      "loss": 0.9375,
+      "step": 5559
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.355426665405607e-06,
+      "loss": 0.8698,
+      "step": 5560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.35231713555276e-06,
+      "loss": 0.8243,
+      "step": 5561
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.349207668588053e-06,
+      "loss": 0.8477,
+      "step": 5562
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.34609826481341e-06,
+      "loss": 0.8235,
+      "step": 5563
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.342988924530742e-06,
+      "loss": 0.8625,
+      "step": 5564
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.33987964804196e-06,
+      "loss": 0.83,
+      "step": 5565
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.336770435648963e-06,
+      "loss": 0.9107,
+      "step": 5566
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.333661287653648e-06,
+      "loss": 0.8442,
+      "step": 5567
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.330552204357904e-06,
+      "loss": 0.8711,
+      "step": 5568
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.32744318606361e-06,
+      "loss": 0.8691,
+      "step": 5569
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.32433423307265e-06,
+      "loss": 0.8583,
+      "step": 5570
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.321225345686893e-06,
+      "loss": 0.8918,
+      "step": 5571
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.318116524208198e-06,
+      "loss": 0.926,
+      "step": 5572
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.315007768938428e-06,
+      "loss": 0.9298,
+      "step": 5573
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.311899080179433e-06,
+      "loss": 0.8717,
+      "step": 5574
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.308790458233055e-06,
+      "loss": 0.8133,
+      "step": 5575
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.305681903401133e-06,
+      "loss": 0.8835,
+      "step": 5576
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.302573415985501e-06,
+      "loss": 0.8809,
+      "step": 5577
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.299464996287984e-06,
+      "loss": 0.8352,
+      "step": 5578
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.296356644610397e-06,
+      "loss": 0.8988,
+      "step": 5579
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.293248361254557e-06,
+      "loss": 0.8587,
+      "step": 5580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.290140146522262e-06,
+      "loss": 0.9048,
+      "step": 5581
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.287032000715318e-06,
+      "loss": 0.8163,
+      "step": 5582
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.283923924135514e-06,
+      "loss": 0.8843,
+      "step": 5583
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.28081591708464e-06,
+      "loss": 0.8833,
+      "step": 5584
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.277707979864466e-06,
+      "loss": 0.8094,
+      "step": 5585
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.27460011277677e-06,
+      "loss": 0.8729,
+      "step": 5586
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.271492316123317e-06,
+      "loss": 0.7724,
+      "step": 5587
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.268384590205858e-06,
+      "loss": 0.8949,
+      "step": 5588
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.265276935326152e-06,
+      "loss": 0.9102,
+      "step": 5589
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.262169351785944e-06,
+      "loss": 0.8958,
+      "step": 5590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.259061839886965e-06,
+      "loss": 0.8466,
+      "step": 5591
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.255954399930948e-06,
+      "loss": 0.8812,
+      "step": 5592
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.252847032219615e-06,
+      "loss": 0.9048,
+      "step": 5593
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.249739737054686e-06,
+      "loss": 0.9441,
+      "step": 5594
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.246632514737873e-06,
+      "loss": 0.9275,
+      "step": 5595
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.24352536557087e-06,
+      "loss": 0.9135,
+      "step": 5596
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.240418289855379e-06,
+      "loss": 0.8846,
+      "step": 5597
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.237311287893086e-06,
+      "loss": 0.7861,
+      "step": 5598
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.23420435998567e-06,
+      "loss": 0.8934,
+      "step": 5599
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.231097506434808e-06,
+      "loss": 0.938,
+      "step": 5600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.227990727542165e-06,
+      "loss": 0.7343,
+      "step": 5601
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.224884023609398e-06,
+      "loss": 0.8415,
+      "step": 5602
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.221777394938162e-06,
+      "loss": 0.7302,
+      "step": 5603
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.218670841830098e-06,
+      "loss": 0.8698,
+      "step": 5604
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.215564364586852e-06,
+      "loss": 0.9414,
+      "step": 5605
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.212457963510045e-06,
+      "loss": 0.8377,
+      "step": 5606
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.209351638901304e-06,
+      "loss": 0.8445,
+      "step": 5607
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.206245391062243e-06,
+      "loss": 0.8592,
+      "step": 5608
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.20313922029447e-06,
+      "loss": 0.8287,
+      "step": 5609
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.200033126899585e-06,
+      "loss": 0.8674,
+      "step": 5610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.196927111179182e-06,
+      "loss": 0.8013,
+      "step": 5611
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.193821173434843e-06,
+      "loss": 0.8736,
+      "step": 5612
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.190715313968147e-06,
+      "loss": 0.8955,
+      "step": 5613
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.187609533080668e-06,
+      "loss": 0.9482,
+      "step": 5614
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.18450383107396e-06,
+      "loss": 0.8467,
+      "step": 5615
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.181398208249583e-06,
+      "loss": 0.8472,
+      "step": 5616
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.178292664909086e-06,
+      "loss": 0.8173,
+      "step": 5617
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.175187201354005e-06,
+      "loss": 0.9247,
+      "step": 5618
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.172081817885876e-06,
+      "loss": 0.8859,
+      "step": 5619
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.168976514806216e-06,
+      "loss": 0.8322,
+      "step": 5620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.165871292416546e-06,
+      "loss": 0.9153,
+      "step": 5621
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.162766151018372e-06,
+      "loss": 0.8964,
+      "step": 5622
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.159661090913195e-06,
+      "loss": 0.9537,
+      "step": 5623
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.156556112402508e-06,
+      "loss": 0.8773,
+      "step": 5624
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.153451215787793e-06,
+      "loss": 0.9537,
+      "step": 5625
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.150346401370528e-06,
+      "loss": 0.8518,
+      "step": 5626
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.147241669452185e-06,
+      "loss": 0.8759,
+      "step": 5627
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.144137020334214e-06,
+      "loss": 0.8951,
+      "step": 5628
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.141032454318082e-06,
+      "loss": 0.863,
+      "step": 5629
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.137927971705223e-06,
+      "loss": 0.8761,
+      "step": 5630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.134823572797076e-06,
+      "loss": 0.8203,
+      "step": 5631
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.131719257895074e-06,
+      "loss": 0.795,
+      "step": 5632
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.12861502730063e-06,
+      "loss": 0.7979,
+      "step": 5633
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.125510881315159e-06,
+      "loss": 0.8205,
+      "step": 5634
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.122406820240064e-06,
+      "loss": 0.9941,
+      "step": 5635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.119302844376741e-06,
+      "loss": 0.818,
+      "step": 5636
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.116198954026577e-06,
+      "loss": 0.9182,
+      "step": 5637
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.113095149490951e-06,
+      "loss": 0.8492,
+      "step": 5638
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.10999143107123e-06,
+      "loss": 0.9176,
+      "step": 5639
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.106887799068782e-06,
+      "loss": 0.8725,
+      "step": 5640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.103784253784958e-06,
+      "loss": 0.9014,
+      "step": 5641
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.100680795521104e-06,
+      "loss": 0.8576,
+      "step": 5642
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.097577424578554e-06,
+      "loss": 0.8579,
+      "step": 5643
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.09447414125864e-06,
+      "loss": 0.9102,
+      "step": 5644
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.09137094586268e-06,
+      "loss": 0.9066,
+      "step": 5645
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.088267838691987e-06,
+      "loss": 0.8604,
+      "step": 5646
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.08516482004786e-06,
+      "loss": 0.8032,
+      "step": 5647
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.0820618902316e-06,
+      "loss": 0.7946,
+      "step": 5648
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.078959049544485e-06,
+      "loss": 0.8564,
+      "step": 5649
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.075856298287796e-06,
+      "loss": 0.9076,
+      "step": 5650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.072753636762795e-06,
+      "loss": 0.8808,
+      "step": 5651
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.069651065270753e-06,
+      "loss": 0.8328,
+      "step": 5652
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.066548584112914e-06,
+      "loss": 0.913,
+      "step": 5653
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.06344619359052e-06,
+      "loss": 0.8347,
+      "step": 5654
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.060343894004805e-06,
+      "loss": 0.951,
+      "step": 5655
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.057241685656995e-06,
+      "loss": 0.8564,
+      "step": 5656
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.054139568848302e-06,
+      "loss": 0.8721,
+      "step": 5657
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.051037543879933e-06,
+      "loss": 0.84,
+      "step": 5658
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.04793561105309e-06,
+      "loss": 0.9419,
+      "step": 5659
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.044833770668957e-06,
+      "loss": 0.8132,
+      "step": 5660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.041732023028714e-06,
+      "loss": 0.8918,
+      "step": 5661
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.038630368433537e-06,
+      "loss": 0.9882,
+      "step": 5662
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.035528807184579e-06,
+      "loss": 0.8957,
+      "step": 5663
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.032427339583e-06,
+      "loss": 0.9283,
+      "step": 5664
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.02932596592994e-06,
+      "loss": 0.8679,
+      "step": 5665
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.026224686526539e-06,
+      "loss": 0.9509,
+      "step": 5666
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.023123501673914e-06,
+      "loss": 0.9511,
+      "step": 5667
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.020022411673186e-06,
+      "loss": 0.8716,
+      "step": 5668
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.016921416825465e-06,
+      "loss": 0.8253,
+      "step": 5669
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.013820517431841e-06,
+      "loss": 0.8098,
+      "step": 5670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.010719713793406e-06,
+      "loss": 0.8242,
+      "step": 5671
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.00761900621124e-06,
+      "loss": 0.8093,
+      "step": 5672
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.004518394986413e-06,
+      "loss": 0.8728,
+      "step": 5673
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.00141788041998e-06,
+      "loss": 0.8677,
+      "step": 5674
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.998317462813e-06,
+      "loss": 0.9131,
+      "step": 5675
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.99521714246651e-06,
+      "loss": 0.8955,
+      "step": 5676
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.992116919681545e-06,
+      "loss": 0.858,
+      "step": 5677
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.989016794759127e-06,
+      "loss": 0.9108,
+      "step": 5678
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.985916768000266e-06,
+      "loss": 0.8893,
+      "step": 5679
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.98281683970597e-06,
+      "loss": 0.9081,
+      "step": 5680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.97971701017723e-06,
+      "loss": 0.889,
+      "step": 5681
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.97661727971503e-06,
+      "loss": 0.8643,
+      "step": 5682
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.973517648620348e-06,
+      "loss": 0.8736,
+      "step": 5683
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.970418117194146e-06,
+      "loss": 0.8016,
+      "step": 5684
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.967318685737385e-06,
+      "loss": 0.8987,
+      "step": 5685
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.964219354550999e-06,
+      "loss": 0.8605,
+      "step": 5686
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.96112012393594e-06,
+      "loss": 0.8313,
+      "step": 5687
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.958020994193124e-06,
+      "loss": 0.864,
+      "step": 5688
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.954921965623471e-06,
+      "loss": 0.8605,
+      "step": 5689
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.951823038527887e-06,
+      "loss": 0.8614,
+      "step": 5690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.94872421320727e-06,
+      "loss": 0.8756,
+      "step": 5691
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.945625489962503e-06,
+      "loss": 0.8195,
+      "step": 5692
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.942526869094468e-06,
+      "loss": 0.8853,
+      "step": 5693
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.93942835090403e-06,
+      "loss": 0.8232,
+      "step": 5694
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.936329935692046e-06,
+      "loss": 0.9027,
+      "step": 5695
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.933231623759365e-06,
+      "loss": 0.7865,
+      "step": 5696
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.930133415406822e-06,
+      "loss": 0.8796,
+      "step": 5697
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.927035310935241e-06,
+      "loss": 0.9612,
+      "step": 5698
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.923937310645446e-06,
+      "loss": 0.8698,
+      "step": 5699
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.920839414838243e-06,
+      "loss": 0.8648,
+      "step": 5700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.917741623814424e-06,
+      "loss": 0.8369,
+      "step": 5701
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.914643937874778e-06,
+      "loss": 0.8822,
+      "step": 5702
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.911546357320085e-06,
+      "loss": 0.9229,
+      "step": 5703
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.908448882451104e-06,
+      "loss": 0.9008,
+      "step": 5704
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.905351513568596e-06,
+      "loss": 0.9111,
+      "step": 5705
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.902254250973306e-06,
+      "loss": 0.9207,
+      "step": 5706
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.899157094965966e-06,
+      "loss": 0.8384,
+      "step": 5707
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.896060045847305e-06,
+      "loss": 0.9031,
+      "step": 5708
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.892963103918034e-06,
+      "loss": 0.8554,
+      "step": 5709
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.88986626947886e-06,
+      "loss": 0.8708,
+      "step": 5710
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.886769542830478e-06,
+      "loss": 0.8619,
+      "step": 5711
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.883672924273566e-06,
+      "loss": 0.9212,
+      "step": 5712
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.880576414108803e-06,
+      "loss": 0.8185,
+      "step": 5713
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.877480012636847e-06,
+      "loss": 0.9493,
+      "step": 5714
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.87438372015835e-06,
+      "loss": 0.8973,
+      "step": 5715
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.871287536973954e-06,
+      "loss": 0.9222,
+      "step": 5716
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.868191463384291e-06,
+      "loss": 0.8921,
+      "step": 5717
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.865095499689978e-06,
+      "loss": 0.882,
+      "step": 5718
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.861999646191623e-06,
+      "loss": 0.8639,
+      "step": 5719
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.85890390318983e-06,
+      "loss": 0.8949,
+      "step": 5720
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.85580827098518e-06,
+      "loss": 0.899,
+      "step": 5721
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.852712749878255e-06,
+      "loss": 0.8929,
+      "step": 5722
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.84961734016962e-06,
+      "loss": 0.9323,
+      "step": 5723
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.846522042159833e-06,
+      "loss": 0.8937,
+      "step": 5724
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.843426856149431e-06,
+      "loss": 0.7901,
+      "step": 5725
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.840331782438954e-06,
+      "loss": 0.8433,
+      "step": 5726
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.837236821328924e-06,
+      "loss": 0.8999,
+      "step": 5727
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.83414197311985e-06,
+      "loss": 0.9147,
+      "step": 5728
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.831047238112235e-06,
+      "loss": 0.8823,
+      "step": 5729
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.82795261660657e-06,
+      "loss": 0.8116,
+      "step": 5730
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.824858108903332e-06,
+      "loss": 0.9258,
+      "step": 5731
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.821763715302986e-06,
+      "loss": 0.7984,
+      "step": 5732
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.818669436105991e-06,
+      "loss": 0.851,
+      "step": 5733
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.815575271612798e-06,
+      "loss": 0.845,
+      "step": 5734
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.812481222123835e-06,
+      "loss": 0.6985,
+      "step": 5735
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.809387287939528e-06,
+      "loss": 0.8827,
+      "step": 5736
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.806293469360287e-06,
+      "loss": 0.8503,
+      "step": 5737
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.803199766686517e-06,
+      "loss": 0.8046,
+      "step": 5738
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.800106180218601e-06,
+      "loss": 0.842,
+      "step": 5739
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.797012710256923e-06,
+      "loss": 0.8583,
+      "step": 5740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.79391935710185e-06,
+      "loss": 0.8191,
+      "step": 5741
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.790826121053732e-06,
+      "loss": 0.9122,
+      "step": 5742
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.787733002412922e-06,
+      "loss": 0.8929,
+      "step": 5743
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.784640001479741e-06,
+      "loss": 0.8771,
+      "step": 5744
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.781547118554523e-06,
+      "loss": 0.877,
+      "step": 5745
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.77845435393757e-06,
+      "loss": 0.9366,
+      "step": 5746
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.775361707929184e-06,
+      "loss": 0.919,
+      "step": 5747
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.772269180829653e-06,
+      "loss": 0.9197,
+      "step": 5748
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.76917677293925e-06,
+      "loss": 0.8535,
+      "step": 5749
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.766084484558237e-06,
+      "loss": 0.886,
+      "step": 5750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.762992315986873e-06,
+      "loss": 0.8869,
+      "step": 5751
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.759900267525393e-06,
+      "loss": 0.9225,
+      "step": 5752
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.756808339474026e-06,
+      "loss": 0.8563,
+      "step": 5753
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.753716532132992e-06,
+      "loss": 0.8881,
+      "step": 5754
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.750624845802494e-06,
+      "loss": 0.8727,
+      "step": 5755
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.747533280782725e-06,
+      "loss": 0.8945,
+      "step": 5756
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.744441837373874e-06,
+      "loss": 0.9208,
+      "step": 5757
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.741350515876103e-06,
+      "loss": 0.8809,
+      "step": 5758
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.738259316589579e-06,
+      "loss": 0.8495,
+      "step": 5759
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.73516823981444e-06,
+      "loss": 0.8878,
+      "step": 5760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.732077285850827e-06,
+      "loss": 0.8749,
+      "step": 5761
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.728986454998858e-06,
+      "loss": 0.8089,
+      "step": 5762
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.725895747558645e-06,
+      "loss": 0.9097,
+      "step": 5763
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.72280516383029e-06,
+      "loss": 0.8843,
+      "step": 5764
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.719714704113875e-06,
+      "loss": 0.9231,
+      "step": 5765
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.716624368709477e-06,
+      "loss": 0.9057,
+      "step": 5766
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.713534157917162e-06,
+      "loss": 0.8223,
+      "step": 5767
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.71044407203697e-06,
+      "loss": 0.8628,
+      "step": 5768
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.707354111368952e-06,
+      "loss": 0.8461,
+      "step": 5769
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.70426427621313e-06,
+      "loss": 0.7822,
+      "step": 5770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.701174566869513e-06,
+      "loss": 0.8795,
+      "step": 5771
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.698084983638111e-06,
+      "loss": 0.8689,
+      "step": 5772
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.69499552681891e-06,
+      "loss": 0.8795,
+      "step": 5773
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.691906196711884e-06,
+      "loss": 0.8342,
+      "step": 5774
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.688816993617004e-06,
+      "loss": 0.8676,
+      "step": 5775
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.685727917834218e-06,
+      "loss": 0.9062,
+      "step": 5776
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.682638969663468e-06,
+      "loss": 0.8916,
+      "step": 5777
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.679550149404685e-06,
+      "loss": 0.9058,
+      "step": 5778
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.676461457357777e-06,
+      "loss": 0.8093,
+      "step": 5779
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.673372893822653e-06,
+      "loss": 0.8793,
+      "step": 5780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.670284459099203e-06,
+      "loss": 0.9203,
+      "step": 5781
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.667196153487308e-06,
+      "loss": 0.8656,
+      "step": 5782
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.664107977286826e-06,
+      "loss": 0.8141,
+      "step": 5783
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.661019930797615e-06,
+      "loss": 0.8655,
+      "step": 5784
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.657932014319517e-06,
+      "loss": 0.8486,
+      "step": 5785
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.654844228152355e-06,
+      "loss": 0.9164,
+      "step": 5786
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.651756572595948e-06,
+      "loss": 0.8667,
+      "step": 5787
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.648669047950097e-06,
+      "loss": 0.8788,
+      "step": 5788
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.64558165451459e-06,
+      "loss": 0.7826,
+      "step": 5789
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.642494392589206e-06,
+      "loss": 0.8715,
+      "step": 5790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.639407262473708e-06,
+      "loss": 0.8305,
+      "step": 5791
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.63632026446785e-06,
+      "loss": 0.8673,
+      "step": 5792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.633233398871372e-06,
+      "loss": 0.9493,
+      "step": 5793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.630146665983993e-06,
+      "loss": 0.822,
+      "step": 5794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.62706006610543e-06,
+      "loss": 0.8336,
+      "step": 5795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.623973599535385e-06,
+      "loss": 0.8929,
+      "step": 5796
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.620887266573542e-06,
+      "loss": 0.8117,
+      "step": 5797
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.617801067519575e-06,
+      "loss": 0.9193,
+      "step": 5798
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.614715002673148e-06,
+      "loss": 0.8839,
+      "step": 5799
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.611629072333905e-06,
+      "loss": 0.8449,
+      "step": 5800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.608543276801482e-06,
+      "loss": 0.9191,
+      "step": 5801
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.605457616375503e-06,
+      "loss": 0.894,
+      "step": 5802
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.602372091355572e-06,
+      "loss": 0.8629,
+      "step": 5803
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.599286702041292e-06,
+      "loss": 0.9241,
+      "step": 5804
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.59620144873224e-06,
+      "loss": 0.9173,
+      "step": 5805
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.593116331727987e-06,
+      "loss": 0.816,
+      "step": 5806
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.590031351328088e-06,
+      "loss": 0.8502,
+      "step": 5807
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.586946507832088e-06,
+      "loss": 0.8018,
+      "step": 5808
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.583861801539515e-06,
+      "loss": 0.8545,
+      "step": 5809
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.580777232749883e-06,
+      "loss": 0.8853,
+      "step": 5810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.577692801762698e-06,
+      "loss": 0.8584,
+      "step": 5811
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.574608508877448e-06,
+      "loss": 0.8101,
+      "step": 5812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.571524354393608e-06,
+      "loss": 0.9231,
+      "step": 5813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.568440338610638e-06,
+      "loss": 0.87,
+      "step": 5814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.565356461827993e-06,
+      "loss": 0.9324,
+      "step": 5815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.562272724345108e-06,
+      "loss": 0.8542,
+      "step": 5816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.559189126461403e-06,
+      "loss": 0.7601,
+      "step": 5817
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.556105668476287e-06,
+      "loss": 0.8653,
+      "step": 5818
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.553022350689151e-06,
+      "loss": 0.837,
+      "step": 5819
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.549939173399385e-06,
+      "loss": 0.8624,
+      "step": 5820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.546856136906348e-06,
+      "loss": 0.9589,
+      "step": 5821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.5437732415094e-06,
+      "loss": 0.7761,
+      "step": 5822
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.540690487507877e-06,
+      "loss": 0.8508,
+      "step": 5823
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.537607875201106e-06,
+      "loss": 0.9255,
+      "step": 5824
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.534525404888402e-06,
+      "loss": 0.8226,
+      "step": 5825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.531443076869058e-06,
+      "loss": 0.8779,
+      "step": 5826
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.528360891442371e-06,
+      "loss": 0.8339,
+      "step": 5827
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.525278848907603e-06,
+      "loss": 0.938,
+      "step": 5828
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.522196949564014e-06,
+      "loss": 0.887,
+      "step": 5829
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.51911519371085e-06,
+      "loss": 0.886,
+      "step": 5830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.516033581647336e-06,
+      "loss": 0.8431,
+      "step": 5831
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.512952113672689e-06,
+      "loss": 0.8309,
+      "step": 5832
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.509870790086115e-06,
+      "loss": 0.8995,
+      "step": 5833
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.506789611186794e-06,
+      "loss": 0.837,
+      "step": 5834
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.503708577273906e-06,
+      "loss": 0.8698,
+      "step": 5835
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.500627688646607e-06,
+      "loss": 0.8535,
+      "step": 5836
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.497546945604045e-06,
+      "loss": 0.7917,
+      "step": 5837
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.494466348445345e-06,
+      "loss": 0.8674,
+      "step": 5838
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.491385897469633e-06,
+      "loss": 0.9167,
+      "step": 5839
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.48830559297601e-06,
+      "loss": 0.867,
+      "step": 5840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.48522543526356e-06,
+      "loss": 0.8063,
+      "step": 5841
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.48214542463136e-06,
+      "loss": 0.7969,
+      "step": 5842
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.479065561378473e-06,
+      "loss": 0.9205,
+      "step": 5843
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.475985845803938e-06,
+      "loss": 0.9073,
+      "step": 5844
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.472906278206794e-06,
+      "loss": 0.7845,
+      "step": 5845
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.469826858886054e-06,
+      "loss": 0.9177,
+      "step": 5846
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.466747588140722e-06,
+      "loss": 0.8925,
+      "step": 5847
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.463668466269785e-06,
+      "loss": 0.8676,
+      "step": 5848
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.460589493572214e-06,
+      "loss": 0.8688,
+      "step": 5849
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.457510670346976e-06,
+      "loss": 0.8096,
+      "step": 5850
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.454431996893013e-06,
+      "loss": 0.903,
+      "step": 5851
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.451353473509254e-06,
+      "loss": 0.8597,
+      "step": 5852
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.448275100494615e-06,
+      "loss": 0.8729,
+      "step": 5853
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.445196878147997e-06,
+      "loss": 0.895,
+      "step": 5854
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.442118806768285e-06,
+      "loss": 0.8765,
+      "step": 5855
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.439040886654354e-06,
+      "loss": 0.853,
+      "step": 5856
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.435963118105062e-06,
+      "loss": 0.8552,
+      "step": 5857
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.432885501419248e-06,
+      "loss": 0.8151,
+      "step": 5858
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.429808036895739e-06,
+      "loss": 0.8218,
+      "step": 5859
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.426730724833354e-06,
+      "loss": 0.843,
+      "step": 5860
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.423653565530882e-06,
+      "loss": 0.8621,
+      "step": 5861
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.420576559287112e-06,
+      "loss": 0.9347,
+      "step": 5862
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.417499706400815e-06,
+      "loss": 0.7433,
+      "step": 5863
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.414423007170742e-06,
+      "loss": 0.8458,
+      "step": 5864
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.411346461895629e-06,
+      "loss": 0.8531,
+      "step": 5865
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.408270070874201e-06,
+      "loss": 0.8951,
+      "step": 5866
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.40519383440517e-06,
+      "loss": 0.9042,
+      "step": 5867
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.402117752787225e-06,
+      "loss": 0.8677,
+      "step": 5868
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.399041826319047e-06,
+      "loss": 0.856,
+      "step": 5869
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.395966055299302e-06,
+      "loss": 0.8494,
+      "step": 5870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.392890440026635e-06,
+      "loss": 0.9082,
+      "step": 5871
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.389814980799679e-06,
+      "loss": 0.9429,
+      "step": 5872
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.386739677917052e-06,
+      "loss": 0.7503,
+      "step": 5873
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.38366453167736e-06,
+      "loss": 0.7876,
+      "step": 5874
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.380589542379191e-06,
+      "loss": 0.8868,
+      "step": 5875
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.377514710321117e-06,
+      "loss": 0.7906,
+      "step": 5876
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.374440035801692e-06,
+      "loss": 0.8463,
+      "step": 5877
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.371365519119463e-06,
+      "loss": 0.9071,
+      "step": 5878
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.368291160572952e-06,
+      "loss": 0.9019,
+      "step": 5879
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.365216960460675e-06,
+      "loss": 0.9058,
+      "step": 5880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.362142919081125e-06,
+      "loss": 0.836,
+      "step": 5881
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.359069036732781e-06,
+      "loss": 0.8728,
+      "step": 5882
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.355995313714114e-06,
+      "loss": 0.798,
+      "step": 5883
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.352921750323562e-06,
+      "loss": 0.9079,
+      "step": 5884
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.349848346859574e-06,
+      "loss": 0.84,
+      "step": 5885
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.346775103620559e-06,
+      "loss": 0.7873,
+      "step": 5886
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.343702020904924e-06,
+      "loss": 0.8482,
+      "step": 5887
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.340629099011057e-06,
+      "loss": 0.8912,
+      "step": 5888
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.337556338237328e-06,
+      "loss": 0.8887,
+      "step": 5889
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.33448373888209e-06,
+      "loss": 0.8632,
+      "step": 5890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.331411301243691e-06,
+      "loss": 0.8438,
+      "step": 5891
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.328339025620449e-06,
+      "loss": 0.8601,
+      "step": 5892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.325266912310675e-06,
+      "loss": 0.9359,
+      "step": 5893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.322194961612668e-06,
+      "loss": 0.8301,
+      "step": 5894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.319123173824697e-06,
+      "loss": 0.8594,
+      "step": 5895
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.316051549245026e-06,
+      "loss": 0.9109,
+      "step": 5896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.312980088171903e-06,
+      "loss": 0.8302,
+      "step": 5897
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.309908790903562e-06,
+      "loss": 0.9138,
+      "step": 5898
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.30683765773821e-06,
+      "loss": 0.8823,
+      "step": 5899
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.303766688974047e-06,
+      "loss": 0.8704,
+      "step": 5900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.300695884909258e-06,
+      "loss": 0.9047,
+      "step": 5901
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.297625245842006e-06,
+      "loss": 0.8821,
+      "step": 5902
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.294554772070443e-06,
+      "loss": 0.9173,
+      "step": 5903
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.291484463892703e-06,
+      "loss": 0.8787,
+      "step": 5904
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.288414321606904e-06,
+      "loss": 0.9082,
+      "step": 5905
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.285344345511147e-06,
+      "loss": 0.8843,
+      "step": 5906
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.282274535903518e-06,
+      "loss": 0.8866,
+      "step": 5907
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.279204893082083e-06,
+      "loss": 0.965,
+      "step": 5908
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.276135417344908e-06,
+      "loss": 0.8818,
+      "step": 5909
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.273066108990017e-06,
+      "loss": 0.8427,
+      "step": 5910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.269996968315437e-06,
+      "loss": 0.8514,
+      "step": 5911
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.266927995619175e-06,
+      "loss": 0.8617,
+      "step": 5912
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.263859191199212e-06,
+      "loss": 0.8413,
+      "step": 5913
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.260790555353526e-06,
+      "loss": 0.9043,
+      "step": 5914
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.257722088380072e-06,
+      "loss": 0.8871,
+      "step": 5915
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.254653790576787e-06,
+      "loss": 0.8231,
+      "step": 5916
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.251585662241595e-06,
+      "loss": 0.8172,
+      "step": 5917
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.248517703672405e-06,
+      "loss": 0.8951,
+      "step": 5918
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.245449915167098e-06,
+      "loss": 0.8804,
+      "step": 5919
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.242382297023558e-06,
+      "loss": 0.8606,
+      "step": 5920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.239314849539639e-06,
+      "loss": 0.8733,
+      "step": 5921
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.23624757301318e-06,
+      "loss": 0.858,
+      "step": 5922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.233180467742003e-06,
+      "loss": 0.809,
+      "step": 5923
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.230113534023917e-06,
+      "loss": 0.9336,
+      "step": 5924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.227046772156716e-06,
+      "loss": 0.8809,
+      "step": 5925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.223980182438167e-06,
+      "loss": 0.8887,
+      "step": 5926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.22091376516603e-06,
+      "loss": 0.9702,
+      "step": 5927
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.217847520638049e-06,
+      "loss": 0.8977,
+      "step": 5928
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.21478144915194e-06,
+      "loss": 0.8922,
+      "step": 5929
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.211715551005414e-06,
+      "loss": 0.9317,
+      "step": 5930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.20864982649616e-06,
+      "loss": 0.9419,
+      "step": 5931
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.205584275921854e-06,
+      "loss": 0.8266,
+      "step": 5932
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.20251889958015e-06,
+      "loss": 0.9065,
+      "step": 5933
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.199453697768686e-06,
+      "loss": 0.856,
+      "step": 5934
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.196388670785085e-06,
+      "loss": 0.8176,
+      "step": 5935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.193323818926955e-06,
+      "loss": 0.8507,
+      "step": 5936
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.190259142491878e-06,
+      "loss": 0.8381,
+      "step": 5937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.187194641777432e-06,
+      "loss": 0.8369,
+      "step": 5938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.184130317081169e-06,
+      "loss": 0.873,
+      "step": 5939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.181066168700622e-06,
+      "loss": 0.8619,
+      "step": 5940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.178002196933318e-06,
+      "loss": 0.9386,
+      "step": 5941
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.174938402076754e-06,
+      "loss": 0.8247,
+      "step": 5942
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.171874784428415e-06,
+      "loss": 0.8324,
+      "step": 5943
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.168811344285776e-06,
+      "loss": 0.8816,
+      "step": 5944
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.165748081946284e-06,
+      "loss": 0.9698,
+      "step": 5945
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.162684997707374e-06,
+      "loss": 0.8624,
+      "step": 5946
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.159622091866463e-06,
+      "loss": 0.9218,
+      "step": 5947
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.156559364720947e-06,
+      "loss": 0.8995,
+      "step": 5948
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.153496816568215e-06,
+      "loss": 0.9135,
+      "step": 5949
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.150434447705623e-06,
+      "loss": 0.9122,
+      "step": 5950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.147372258430524e-06,
+      "loss": 0.8697,
+      "step": 5951
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.144310249040246e-06,
+      "loss": 0.9031,
+      "step": 5952
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.141248419832101e-06,
+      "loss": 0.8855,
+      "step": 5953
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.138186771103382e-06,
+      "loss": 0.8066,
+      "step": 5954
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.13512530315137e-06,
+      "loss": 0.7868,
+      "step": 5955
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.132064016273325e-06,
+      "loss": 0.8915,
+      "step": 5956
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.129002910766486e-06,
+      "loss": 0.897,
+      "step": 5957
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.12594198692808e-06,
+      "loss": 0.8294,
+      "step": 5958
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.122881245055315e-06,
+      "loss": 0.84,
+      "step": 5959
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.119820685445373e-06,
+      "loss": 0.8398,
+      "step": 5960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.116760308395433e-06,
+      "loss": 0.8624,
+      "step": 5961
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.113700114202647e-06,
+      "loss": 0.8542,
+      "step": 5962
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.110640103164148e-06,
+      "loss": 0.9182,
+      "step": 5963
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.107580275577059e-06,
+      "loss": 0.9313,
+      "step": 5964
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.104520631738478e-06,
+      "loss": 0.8558,
+      "step": 5965
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.101461171945483e-06,
+      "loss": 0.9214,
+      "step": 5966
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.09840189649515e-06,
+      "loss": 0.8203,
+      "step": 5967
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.095342805684516e-06,
+      "loss": 0.8634,
+      "step": 5968
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.092283899810612e-06,
+      "loss": 0.8539,
+      "step": 5969
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.089225179170454e-06,
+      "loss": 0.8647,
+      "step": 5970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.086166644061028e-06,
+      "loss": 0.8609,
+      "step": 5971
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.083108294779313e-06,
+      "loss": 0.8069,
+      "step": 5972
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.080050131622268e-06,
+      "loss": 0.8465,
+      "step": 5973
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.076992154886826e-06,
+      "loss": 0.8684,
+      "step": 5974
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.07393436486991e-06,
+      "loss": 0.8516,
+      "step": 5975
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.070876761868426e-06,
+      "loss": 0.8844,
+      "step": 5976
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.067819346179253e-06,
+      "loss": 0.957,
+      "step": 5977
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.064762118099258e-06,
+      "loss": 0.8667,
+      "step": 5978
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.061705077925293e-06,
+      "loss": 0.9188,
+      "step": 5979
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.058648225954188e-06,
+      "loss": 0.8851,
+      "step": 5980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.05559156248275e-06,
+      "loss": 0.9142,
+      "step": 5981
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.052535087807774e-06,
+      "loss": 0.8335,
+      "step": 5982
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.049478802226039e-06,
+      "loss": 0.9036,
+      "step": 5983
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.046422706034294e-06,
+      "loss": 0.9167,
+      "step": 5984
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.043366799529282e-06,
+      "loss": 0.9,
+      "step": 5985
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.040311083007725e-06,
+      "loss": 0.8077,
+      "step": 5986
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.037255556766317e-06,
+      "loss": 0.8799,
+      "step": 5987
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.034200221101746e-06,
+      "loss": 0.8622,
+      "step": 5988
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.031145076310672e-06,
+      "loss": 0.8387,
+      "step": 5989
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.028090122689747e-06,
+      "loss": 0.8378,
+      "step": 5990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.025035360535597e-06,
+      "loss": 0.8558,
+      "step": 5991
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.021980790144828e-06,
+      "loss": 0.8535,
+      "step": 5992
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.018926411814029e-06,
+      "loss": 0.7911,
+      "step": 5993
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.015872225839776e-06,
+      "loss": 0.8697,
+      "step": 5994
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.012818232518616e-06,
+      "loss": 0.7776,
+      "step": 5995
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.009764432147086e-06,
+      "loss": 0.8313,
+      "step": 5996
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.006710825021703e-06,
+      "loss": 0.7736,
+      "step": 5997
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.003657411438961e-06,
+      "loss": 0.8286,
+      "step": 5998
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.000604191695337e-06,
+      "loss": 0.8998,
+      "step": 5999
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.997551166087293e-06,
+      "loss": 0.8165,
+      "step": 6000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.994498334911261e-06,
+      "loss": 0.8947,
+      "step": 6001
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.991445698463672e-06,
+      "loss": 0.9624,
+      "step": 6002
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.988393257040925e-06,
+      "loss": 0.8607,
+      "step": 6003
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.985341010939402e-06,
+      "loss": 0.8763,
+      "step": 6004
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.982288960455469e-06,
+      "loss": 0.8207,
+      "step": 6005
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.979237105885467e-06,
+      "loss": 0.8205,
+      "step": 6006
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.97618544752573e-06,
+      "loss": 0.8596,
+      "step": 6007
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.973133985672558e-06,
+      "loss": 0.9054,
+      "step": 6008
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.97008272062224e-06,
+      "loss": 0.8811,
+      "step": 6009
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.967031652671051e-06,
+      "loss": 0.8415,
+      "step": 6010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.963980782115235e-06,
+      "loss": 0.9064,
+      "step": 6011
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.960930109251023e-06,
+      "loss": 0.9054,
+      "step": 6012
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.957879634374628e-06,
+      "loss": 0.887,
+      "step": 6013
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.954829357782243e-06,
+      "loss": 0.8949,
+      "step": 6014
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.951779279770044e-06,
+      "loss": 0.8651,
+      "step": 6015
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.948729400634178e-06,
+      "loss": 0.8797,
+      "step": 6016
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.945679720670783e-06,
+      "loss": 0.8782,
+      "step": 6017
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.942630240175977e-06,
+      "loss": 0.963,
+      "step": 6018
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.939580959445849e-06,
+      "loss": 0.9332,
+      "step": 6019
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.936531878776484e-06,
+      "loss": 0.9187,
+      "step": 6020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.933482998463929e-06,
+      "loss": 0.8817,
+      "step": 6021
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.930434318804229e-06,
+      "loss": 0.9165,
+      "step": 6022
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.9273858400934e-06,
+      "loss": 0.8882,
+      "step": 6023
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.924337562627435e-06,
+      "loss": 0.8117,
+      "step": 6024
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.921289486702324e-06,
+      "loss": 0.8,
+      "step": 6025
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.918241612614016e-06,
+      "loss": 0.8525,
+      "step": 6026
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.91519394065846e-06,
+      "loss": 0.8672,
+      "step": 6027
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.91214647113157e-06,
+      "loss": 0.915,
+      "step": 6028
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.909099204329247e-06,
+      "loss": 0.9142,
+      "step": 6029
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.906052140547373e-06,
+      "loss": 0.8621,
+      "step": 6030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.90300528008181e-06,
+      "loss": 0.8636,
+      "step": 6031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.899958623228398e-06,
+      "loss": 0.8705,
+      "step": 6032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.896912170282957e-06,
+      "loss": 0.8472,
+      "step": 6033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.893865921541294e-06,
+      "loss": 0.8606,
+      "step": 6034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.890819877299185e-06,
+      "loss": 0.7572,
+      "step": 6035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.887774037852395e-06,
+      "loss": 0.869,
+      "step": 6036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.884728403496666e-06,
+      "loss": 0.9079,
+      "step": 6037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.881682974527723e-06,
+      "loss": 0.809,
+      "step": 6038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.878637751241263e-06,
+      "loss": 0.9262,
+      "step": 6039
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.875592733932972e-06,
+      "loss": 0.9515,
+      "step": 6040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.872547922898515e-06,
+      "loss": 0.8799,
+      "step": 6041
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.869503318433529e-06,
+      "loss": 0.9171,
+      "step": 6042
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.866458920833635e-06,
+      "loss": 0.8461,
+      "step": 6043
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.863414730394444e-06,
+      "loss": 0.9346,
+      "step": 6044
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.860370747411528e-06,
+      "loss": 0.9346,
+      "step": 6045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.857326972180455e-06,
+      "loss": 0.838,
+      "step": 6046
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.854283404996768e-06,
+      "loss": 0.9337,
+      "step": 6047
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.85124004615598e-06,
+      "loss": 0.8427,
+      "step": 6048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.848196895953604e-06,
+      "loss": 0.8792,
+      "step": 6049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.845153954685114e-06,
+      "loss": 0.9126,
+      "step": 6050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.842111222645974e-06,
+      "loss": 0.8339,
+      "step": 6051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.839068700131623e-06,
+      "loss": 0.8665,
+      "step": 6052
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.836026387437482e-06,
+      "loss": 0.913,
+      "step": 6053
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.832984284858946e-06,
+      "loss": 0.7899,
+      "step": 6054
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.829942392691402e-06,
+      "loss": 0.847,
+      "step": 6055
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.826900711230204e-06,
+      "loss": 0.8596,
+      "step": 6056
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.823859240770693e-06,
+      "loss": 0.8964,
+      "step": 6057
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.820817981608185e-06,
+      "loss": 0.8705,
+      "step": 6058
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.817776934037975e-06,
+      "loss": 0.8103,
+      "step": 6059
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.814736098355348e-06,
+      "loss": 0.788,
+      "step": 6060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.811695474855554e-06,
+      "loss": 0.9097,
+      "step": 6061
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.808655063833832e-06,
+      "loss": 0.8362,
+      "step": 6062
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.805614865585397e-06,
+      "loss": 0.8652,
+      "step": 6063
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.802574880405438e-06,
+      "loss": 0.8673,
+      "step": 6064
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.799535108589137e-06,
+      "loss": 0.8917,
+      "step": 6065
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.79649555043164e-06,
+      "loss": 0.7815,
+      "step": 6066
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.793456206228085e-06,
+      "loss": 0.8652,
+      "step": 6067
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.790417076273581e-06,
+      "loss": 0.9116,
+      "step": 6068
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.787378160863218e-06,
+      "loss": 0.8022,
+      "step": 6069
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.784339460292065e-06,
+      "loss": 0.874,
+      "step": 6070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.781300974855172e-06,
+      "loss": 0.8235,
+      "step": 6071
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.77826270484757e-06,
+      "loss": 0.8107,
+      "step": 6072
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.775224650564267e-06,
+      "loss": 0.8115,
+      "step": 6073
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.772186812300244e-06,
+      "loss": 0.8518,
+      "step": 6074
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.769149190350469e-06,
+      "loss": 0.8695,
+      "step": 6075
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.766111785009888e-06,
+      "loss": 0.8951,
+      "step": 6076
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.763074596573422e-06,
+      "loss": 0.8443,
+      "step": 6077
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.760037625335973e-06,
+      "loss": 0.8258,
+      "step": 6078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.757000871592426e-06,
+      "loss": 0.9495,
+      "step": 6079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.753964335637634e-06,
+      "loss": 0.8625,
+      "step": 6080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.750928017766444e-06,
+      "loss": 0.8629,
+      "step": 6081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.747891918273668e-06,
+      "loss": 0.9094,
+      "step": 6082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.7448560374541e-06,
+      "loss": 0.9753,
+      "step": 6083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.741820375602524e-06,
+      "loss": 0.8363,
+      "step": 6084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.738784933013687e-06,
+      "loss": 0.9429,
+      "step": 6085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.735749709982329e-06,
+      "loss": 0.8986,
+      "step": 6086
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.732714706803154e-06,
+      "loss": 0.8969,
+      "step": 6087
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.729679923770855e-06,
+      "loss": 0.876,
+      "step": 6088
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.7266453611801e-06,
+      "loss": 0.8819,
+      "step": 6089
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.723611019325538e-06,
+      "loss": 0.8523,
+      "step": 6090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.720576898501793e-06,
+      "loss": 0.9244,
+      "step": 6091
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.71754299900347e-06,
+      "loss": 0.8815,
+      "step": 6092
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.714509321125152e-06,
+      "loss": 0.9081,
+      "step": 6093
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.7114758651614e-06,
+      "loss": 0.8955,
+      "step": 6094
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.708442631406754e-06,
+      "loss": 0.8186,
+      "step": 6095
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.705409620155733e-06,
+      "loss": 0.8444,
+      "step": 6096
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.702376831702834e-06,
+      "loss": 0.8828,
+      "step": 6097
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.699344266342529e-06,
+      "loss": 0.8694,
+      "step": 6098
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.696311924369278e-06,
+      "loss": 0.9128,
+      "step": 6099
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.693279806077504e-06,
+      "loss": 0.8516,
+      "step": 6100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.690247911761621e-06,
+      "loss": 0.8881,
+      "step": 6101
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.68721624171602e-06,
+      "loss": 0.8641,
+      "step": 6102
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.684184796235063e-06,
+      "loss": 0.8255,
+      "step": 6103
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.681153575613098e-06,
+      "loss": 0.8983,
+      "step": 6104
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.678122580144447e-06,
+      "loss": 0.8833,
+      "step": 6105
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.675091810123404e-06,
+      "loss": 0.8502,
+      "step": 6106
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.672061265844259e-06,
+      "loss": 0.861,
+      "step": 6107
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.669030947601265e-06,
+      "loss": 0.8928,
+      "step": 6108
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.666000855688656e-06,
+      "loss": 0.8193,
+      "step": 6109
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.662970990400647e-06,
+      "loss": 0.8783,
+      "step": 6110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.659941352031424e-06,
+      "loss": 0.817,
+      "step": 6111
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.656911940875163e-06,
+      "loss": 0.8636,
+      "step": 6112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.65388275722601e-06,
+      "loss": 0.797,
+      "step": 6113
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.650853801378084e-06,
+      "loss": 0.8701,
+      "step": 6114
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.647825073625495e-06,
+      "loss": 0.8499,
+      "step": 6115
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.644796574262322e-06,
+      "loss": 0.8981,
+      "step": 6116
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.64176830358262e-06,
+      "loss": 0.8579,
+      "step": 6117
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.638740261880423e-06,
+      "loss": 0.842,
+      "step": 6118
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.635712449449756e-06,
+      "loss": 0.9072,
+      "step": 6119
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.632684866584606e-06,
+      "loss": 0.8831,
+      "step": 6120
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.629657513578939e-06,
+      "loss": 0.9032,
+      "step": 6121
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.626630390726704e-06,
+      "loss": 0.85,
+      "step": 6122
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.62360349832183e-06,
+      "loss": 0.8812,
+      "step": 6123
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.620576836658212e-06,
+      "loss": 0.8959,
+      "step": 6124
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6175504060297355e-06,
+      "loss": 0.8864,
+      "step": 6125
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.61452420673026e-06,
+      "loss": 0.8111,
+      "step": 6126
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.611498239053614e-06,
+      "loss": 0.925,
+      "step": 6127
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.608472503293615e-06,
+      "loss": 0.8215,
+      "step": 6128
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.605446999744048e-06,
+      "loss": 0.8729,
+      "step": 6129
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.60242172869869e-06,
+      "loss": 0.8371,
+      "step": 6130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5993966904512815e-06,
+      "loss": 0.8895,
+      "step": 6131
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.596371885295542e-06,
+      "loss": 0.8249,
+      "step": 6132
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.593347313525175e-06,
+      "loss": 0.9263,
+      "step": 6133
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.590322975433857e-06,
+      "loss": 0.853,
+      "step": 6134
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.587298871315242e-06,
+      "loss": 0.871,
+      "step": 6135
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.584275001462961e-06,
+      "loss": 0.8151,
+      "step": 6136
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.581251366170625e-06,
+      "loss": 0.9117,
+      "step": 6137
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.578227965731819e-06,
+      "loss": 0.8382,
+      "step": 6138
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.575204800440108e-06,
+      "loss": 0.8041,
+      "step": 6139
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.572181870589028e-06,
+      "loss": 0.8598,
+      "step": 6140
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.569159176472099e-06,
+      "loss": 0.8028,
+      "step": 6141
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.566136718382821e-06,
+      "loss": 0.8661,
+      "step": 6142
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.563114496614659e-06,
+      "loss": 0.9073,
+      "step": 6143
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.560092511461069e-06,
+      "loss": 0.8643,
+      "step": 6144
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.557070763215471e-06,
+      "loss": 0.7969,
+      "step": 6145
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.55404925217127e-06,
+      "loss": 0.9061,
+      "step": 6146
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.551027978621847e-06,
+      "loss": 0.8776,
+      "step": 6147
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.548006942860557e-06,
+      "loss": 0.7908,
+      "step": 6148
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.544986145180736e-06,
+      "loss": 0.9112,
+      "step": 6149
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.541965585875695e-06,
+      "loss": 0.8459,
+      "step": 6150
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5389452652387186e-06,
+      "loss": 0.858,
+      "step": 6151
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.535925183563073e-06,
+      "loss": 0.8755,
+      "step": 6152
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5329053411419976e-06,
+      "loss": 0.9277,
+      "step": 6153
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.529885738268714e-06,
+      "loss": 0.8254,
+      "step": 6154
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.526866375236417e-06,
+      "loss": 0.8584,
+      "step": 6155
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.523847252338274e-06,
+      "loss": 0.8945,
+      "step": 6156
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.520828369867439e-06,
+      "loss": 0.8253,
+      "step": 6157
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.51780972811703e-06,
+      "loss": 0.8678,
+      "step": 6158
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5147913273801535e-06,
+      "loss": 0.7495,
+      "step": 6159
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.511773167949885e-06,
+      "loss": 0.8792,
+      "step": 6160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.508755250119279e-06,
+      "loss": 0.8801,
+      "step": 6161
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.5057375741813685e-06,
+      "loss": 0.8323,
+      "step": 6162
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.50272014042916e-06,
+      "loss": 0.8415,
+      "step": 6163
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.499702949155634e-06,
+      "loss": 0.8139,
+      "step": 6164
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.496686000653759e-06,
+      "loss": 0.8708,
+      "step": 6165
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.493669295216467e-06,
+      "loss": 0.8951,
+      "step": 6166
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.490652833136671e-06,
+      "loss": 0.8932,
+      "step": 6167
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.487636614707265e-06,
+      "loss": 0.8103,
+      "step": 6168
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.484620640221109e-06,
+      "loss": 0.8831,
+      "step": 6169
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.48160490997105e-06,
+      "loss": 0.8765,
+      "step": 6170
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4785894242499065e-06,
+      "loss": 0.8978,
+      "step": 6171
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.475574183350471e-06,
+      "loss": 0.8574,
+      "step": 6172
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.472559187565515e-06,
+      "loss": 0.8293,
+      "step": 6173
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.46954443718779e-06,
+      "loss": 0.8636,
+      "step": 6174
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.466529932510014e-06,
+      "loss": 0.9255,
+      "step": 6175
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.463515673824888e-06,
+      "loss": 0.8396,
+      "step": 6176
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.46050166142509e-06,
+      "loss": 0.79,
+      "step": 6177
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.457487895603273e-06,
+      "loss": 0.847,
+      "step": 6178
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.454474376652061e-06,
+      "loss": 0.9029,
+      "step": 6179
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.451461104864061e-06,
+      "loss": 0.7957,
+      "step": 6180
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.448448080531854e-06,
+      "loss": 0.8837,
+      "step": 6181
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.44543530394799e-06,
+      "loss": 0.8394,
+      "step": 6182
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.442422775405007e-06,
+      "loss": 0.8475,
+      "step": 6183
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.439410495195411e-06,
+      "loss": 0.7964,
+      "step": 6184
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4363984636116845e-06,
+      "loss": 0.8616,
+      "step": 6185
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.433386680946288e-06,
+      "loss": 0.9343,
+      "step": 6186
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.43037514749166e-06,
+      "loss": 0.8687,
+      "step": 6187
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.427363863540202e-06,
+      "loss": 0.89,
+      "step": 6188
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4243528293843136e-06,
+      "loss": 0.8785,
+      "step": 6189
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.421342045316351e-06,
+      "loss": 0.9389,
+      "step": 6190
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.418331511628651e-06,
+      "loss": 0.8114,
+      "step": 6191
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.415321228613534e-06,
+      "loss": 0.8366,
+      "step": 6192
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.412311196563284e-06,
+      "loss": 0.8232,
+      "step": 6193
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.409301415770168e-06,
+      "loss": 0.8752,
+      "step": 6194
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4062918865264286e-06,
+      "loss": 0.806,
+      "step": 6195
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.403282609124281e-06,
+      "loss": 0.8454,
+      "step": 6196
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.400273583855917e-06,
+      "loss": 0.9122,
+      "step": 6197
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.397264811013507e-06,
+      "loss": 0.9143,
+      "step": 6198
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.39425629088919e-06,
+      "loss": 0.8592,
+      "step": 6199
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.391248023775084e-06,
+      "loss": 0.8881,
+      "step": 6200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.38824000996329e-06,
+      "loss": 0.8867,
+      "step": 6201
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.385232249745873e-06,
+      "loss": 0.9092,
+      "step": 6202
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3822247434148775e-06,
+      "loss": 0.811,
+      "step": 6203
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.379217491262325e-06,
+      "loss": 0.8892,
+      "step": 6204
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.376210493580211e-06,
+      "loss": 0.8672,
+      "step": 6205
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.373203750660505e-06,
+      "loss": 0.8183,
+      "step": 6206
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.370197262795154e-06,
+      "loss": 0.8864,
+      "step": 6207
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.36719103027608e-06,
+      "loss": 0.871,
+      "step": 6208
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.364185053395176e-06,
+      "loss": 0.7989,
+      "step": 6209
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.361179332444318e-06,
+      "loss": 0.8875,
+      "step": 6210
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3581738677153484e-06,
+      "loss": 0.9453,
+      "step": 6211
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.355168659500094e-06,
+      "loss": 0.8557,
+      "step": 6212
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.352163708090353e-06,
+      "loss": 0.8505,
+      "step": 6213
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3491590137778915e-06,
+      "loss": 0.9178,
+      "step": 6214
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.346154576854461e-06,
+      "loss": 0.8776,
+      "step": 6215
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.343150397611782e-06,
+      "loss": 0.8461,
+      "step": 6216
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.340146476341552e-06,
+      "loss": 0.8557,
+      "step": 6217
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3371428133354435e-06,
+      "loss": 0.8882,
+      "step": 6218
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.334139408885102e-06,
+      "loss": 0.8564,
+      "step": 6219
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.33113626328215e-06,
+      "loss": 0.8426,
+      "step": 6220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.328133376818187e-06,
+      "loss": 0.8602,
+      "step": 6221
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.325130749784781e-06,
+      "loss": 0.8412,
+      "step": 6222
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.322128382473477e-06,
+      "loss": 0.8989,
+      "step": 6223
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3191262751758005e-06,
+      "loss": 0.9423,
+      "step": 6224
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.316124428183247e-06,
+      "loss": 0.8143,
+      "step": 6225
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3131228417872905e-06,
+      "loss": 0.8957,
+      "step": 6226
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.310121516279367e-06,
+      "loss": 0.8517,
+      "step": 6227
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.307120451950902e-06,
+      "loss": 0.8721,
+      "step": 6228
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.304119649093293e-06,
+      "loss": 0.86,
+      "step": 6229
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.301119107997905e-06,
+      "loss": 0.8428,
+      "step": 6230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.298118828956081e-06,
+      "loss": 0.865,
+      "step": 6231
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.295118812259145e-06,
+      "loss": 0.8766,
+      "step": 6232
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.292119058198384e-06,
+      "loss": 0.9032,
+      "step": 6233
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.289119567065068e-06,
+      "loss": 0.9111,
+      "step": 6234
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.286120339150438e-06,
+      "loss": 0.8879,
+      "step": 6235
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2831213747457155e-06,
+      "loss": 0.8848,
+      "step": 6236
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2801226741420845e-06,
+      "loss": 0.8391,
+      "step": 6237
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2771242376307125e-06,
+      "loss": 0.8825,
+      "step": 6238
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.274126065502742e-06,
+      "loss": 0.9442,
+      "step": 6239
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.271128158049283e-06,
+      "loss": 0.902,
+      "step": 6240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.268130515561424e-06,
+      "loss": 0.8641,
+      "step": 6241
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2651331383302326e-06,
+      "loss": 0.8349,
+      "step": 6242
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.262136026646738e-06,
+      "loss": 0.9044,
+      "step": 6243
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2591391808019555e-06,
+      "loss": 0.8794,
+      "step": 6244
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2561426010868706e-06,
+      "loss": 0.8577,
+      "step": 6245
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.253146287792434e-06,
+      "loss": 0.9387,
+      "step": 6246
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.250150241209594e-06,
+      "loss": 0.8035,
+      "step": 6247
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.247154461629248e-06,
+      "loss": 0.9185,
+      "step": 6248
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.244158949342279e-06,
+      "loss": 0.8869,
+      "step": 6249
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.241163704639547e-06,
+      "loss": 0.8872,
+      "step": 6250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.238168727811876e-06,
+      "loss": 0.8221,
+      "step": 6251
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.235174019150071e-06,
+      "loss": 0.8668,
+      "step": 6252
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.232179578944914e-06,
+      "loss": 0.8017,
+      "step": 6253
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.229185407487149e-06,
+      "loss": 0.8026,
+      "step": 6254
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.226191505067508e-06,
+      "loss": 0.8678,
+      "step": 6255
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2231978719766884e-06,
+      "loss": 0.8105,
+      "step": 6256
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.22020450850536e-06,
+      "loss": 0.8322,
+      "step": 6257
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.217211414944171e-06,
+      "loss": 0.821,
+      "step": 6258
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.214218591583746e-06,
+      "loss": 0.9506,
+      "step": 6259
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2112260387146784e-06,
+      "loss": 0.856,
+      "step": 6260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.208233756627534e-06,
+      "loss": 0.865,
+      "step": 6261
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2052417456128565e-06,
+      "loss": 0.8251,
+      "step": 6262
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2022500059611625e-06,
+      "loss": 0.8757,
+      "step": 6263
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.199258537962936e-06,
+      "loss": 0.8334,
+      "step": 6264
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.196267341908647e-06,
+      "loss": 0.8175,
+      "step": 6265
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.193276418088729e-06,
+      "loss": 0.8677,
+      "step": 6266
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.190285766793591e-06,
+      "loss": 0.878,
+      "step": 6267
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.187295388313618e-06,
+      "loss": 0.8724,
+      "step": 6268
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.184305282939167e-06,
+      "loss": 0.9047,
+      "step": 6269
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.181315450960562e-06,
+      "loss": 0.9233,
+      "step": 6270
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.178325892668121e-06,
+      "loss": 0.8456,
+      "step": 6271
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.175336608352113e-06,
+      "loss": 0.8611,
+      "step": 6272
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.172347598302787e-06,
+      "loss": 0.882,
+      "step": 6273
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.169358862810374e-06,
+      "loss": 0.8384,
+      "step": 6274
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.166370402165065e-06,
+      "loss": 0.9527,
+      "step": 6275
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.163382216657033e-06,
+      "loss": 0.9015,
+      "step": 6276
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.160394306576426e-06,
+      "loss": 0.8476,
+      "step": 6277
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1574066722133565e-06,
+      "loss": 0.8487,
+      "step": 6278
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.154419313857919e-06,
+      "loss": 0.8553,
+      "step": 6279
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.151432231800173e-06,
+      "loss": 0.884,
+      "step": 6280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.148445426330155e-06,
+      "loss": 0.8686,
+      "step": 6281
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.145458897737882e-06,
+      "loss": 0.8635,
+      "step": 6282
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.142472646313332e-06,
+      "loss": 0.9598,
+      "step": 6283
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.139486672346466e-06,
+      "loss": 0.837,
+      "step": 6284
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.136500976127208e-06,
+      "loss": 0.8915,
+      "step": 6285
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.133515557945463e-06,
+      "loss": 0.8276,
+      "step": 6286
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.130530418091108e-06,
+      "loss": 0.8779,
+      "step": 6287
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.12754555685399e-06,
+      "loss": 0.8875,
+      "step": 6288
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1245609745239285e-06,
+      "loss": 0.8633,
+      "step": 6289
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.121576671390722e-06,
+      "loss": 0.8652,
+      "step": 6290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.118592647744134e-06,
+      "loss": 0.831,
+      "step": 6291
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.115608903873905e-06,
+      "loss": 0.7934,
+      "step": 6292
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.112625440069747e-06,
+      "loss": 0.8821,
+      "step": 6293
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.109642256621353e-06,
+      "loss": 0.9313,
+      "step": 6294
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.106659353818374e-06,
+      "loss": 0.8628,
+      "step": 6295
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.103676731950443e-06,
+      "loss": 0.8931,
+      "step": 6296
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1006943913071665e-06,
+      "loss": 0.8831,
+      "step": 6297
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0977123321781176e-06,
+      "loss": 0.7637,
+      "step": 6298
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0947305548528475e-06,
+      "loss": 0.8645,
+      "step": 6299
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.091749059620881e-06,
+      "loss": 0.9145,
+      "step": 6300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.088767846771706e-06,
+      "loss": 0.9211,
+      "step": 6301
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0857869165947945e-06,
+      "loss": 0.8189,
+      "step": 6302
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.082806269379586e-06,
+      "loss": 0.8727,
+      "step": 6303
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.079825905415491e-06,
+      "loss": 0.8084,
+      "step": 6304
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.076845824991892e-06,
+      "loss": 0.8739,
+      "step": 6305
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.073866028398153e-06,
+      "loss": 0.8375,
+      "step": 6306
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.070886515923599e-06,
+      "loss": 0.9481,
+      "step": 6307
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.067907287857535e-06,
+      "loss": 0.8457,
+      "step": 6308
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.064928344489233e-06,
+      "loss": 0.8615,
+      "step": 6309
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.061949686107938e-06,
+      "loss": 0.8778,
+      "step": 6310
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.058971313002876e-06,
+      "loss": 0.8936,
+      "step": 6311
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0559932254632315e-06,
+      "loss": 0.8816,
+      "step": 6312
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.05301542377817e-06,
+      "loss": 0.845,
+      "step": 6313
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0500379082368305e-06,
+      "loss": 0.7662,
+      "step": 6314
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.047060679128317e-06,
+      "loss": 0.842,
+      "step": 6315
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.044083736741711e-06,
+      "loss": 0.8379,
+      "step": 6316
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0411070813660664e-06,
+      "loss": 0.8257,
+      "step": 6317
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.03813071329041e-06,
+      "loss": 0.85,
+      "step": 6318
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.035154632803733e-06,
+      "loss": 0.8809,
+      "step": 6319
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.032178840195009e-06,
+      "loss": 0.8764,
+      "step": 6320
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.029203335753178e-06,
+      "loss": 0.8951,
+      "step": 6321
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.026228119767149e-06,
+      "loss": 0.8121,
+      "step": 6322
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0232531925258116e-06,
+      "loss": 0.8632,
+      "step": 6323
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.020278554318023e-06,
+      "loss": 0.851,
+      "step": 6324
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0173042054326064e-06,
+      "loss": 0.9023,
+      "step": 6325
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.014330146158367e-06,
+      "loss": 0.8482,
+      "step": 6326
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0113563767840785e-06,
+      "loss": 0.8132,
+      "step": 6327
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.008382897598477e-06,
+      "loss": 0.8531,
+      "step": 6328
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.005409708890292e-06,
+      "loss": 0.8958,
+      "step": 6329
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.002436810948201e-06,
+      "loss": 0.8769,
+      "step": 6330
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9994642040608685e-06,
+      "loss": 0.8309,
+      "step": 6331
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.996491888516927e-06,
+      "loss": 0.8548,
+      "step": 6332
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9935198646049764e-06,
+      "loss": 0.8996,
+      "step": 6333
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.990548132613592e-06,
+      "loss": 0.9484,
+      "step": 6334
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.987576692831323e-06,
+      "loss": 0.7883,
+      "step": 6335
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.984605545546686e-06,
+      "loss": 0.8186,
+      "step": 6336
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9816346910481716e-06,
+      "loss": 0.7794,
+      "step": 6337
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.978664129624241e-06,
+      "loss": 0.8504,
+      "step": 6338
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.975693861563325e-06,
+      "loss": 0.8671,
+      "step": 6339
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.972723887153828e-06,
+      "loss": 0.7877,
+      "step": 6340
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.96975420668413e-06,
+      "loss": 0.8696,
+      "step": 6341
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9667848204425785e-06,
+      "loss": 0.8288,
+      "step": 6342
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.963815728717488e-06,
+      "loss": 0.8304,
+      "step": 6343
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.960846931797152e-06,
+      "loss": 0.811,
+      "step": 6344
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.957878429969831e-06,
+      "loss": 0.8607,
+      "step": 6345
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9549102235237565e-06,
+      "loss": 0.908,
+      "step": 6346
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.951942312747135e-06,
+      "loss": 0.893,
+      "step": 6347
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.948974697928144e-06,
+      "loss": 0.8728,
+      "step": 6348
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9460073793549244e-06,
+      "loss": 0.882,
+      "step": 6349
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.943040357315598e-06,
+      "loss": 0.9351,
+      "step": 6350
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.940073632098252e-06,
+      "loss": 0.9341,
+      "step": 6351
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9371072039909515e-06,
+      "loss": 0.9185,
+      "step": 6352
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.934141073281726e-06,
+      "loss": 0.9211,
+      "step": 6353
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.931175240258576e-06,
+      "loss": 0.7922,
+      "step": 6354
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.928209705209479e-06,
+      "loss": 0.8025,
+      "step": 6355
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9252444684223765e-06,
+      "loss": 0.9526,
+      "step": 6356
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.922279530185184e-06,
+      "loss": 0.8584,
+      "step": 6357
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.919314890785793e-06,
+      "loss": 0.8486,
+      "step": 6358
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.916350550512057e-06,
+      "loss": 0.8777,
+      "step": 6359
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.913386509651807e-06,
+      "loss": 0.8599,
+      "step": 6360
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.910422768492844e-06,
+      "loss": 0.8218,
+      "step": 6361
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.907459327322934e-06,
+      "loss": 0.8695,
+      "step": 6362
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.904496186429821e-06,
+      "loss": 0.8763,
+      "step": 6363
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.90153334610122e-06,
+      "loss": 0.9255,
+      "step": 6364
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.8985708066248135e-06,
+      "loss": 0.8334,
+      "step": 6365
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.895608568288255e-06,
+      "loss": 0.8934,
+      "step": 6366
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.892646631379168e-06,
+      "loss": 0.8786,
+      "step": 6367
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.889684996185148e-06,
+      "loss": 0.9164,
+      "step": 6368
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.886723662993764e-06,
+      "loss": 0.8827,
+      "step": 6369
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.88376263209255e-06,
+      "loss": 0.9042,
+      "step": 6370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8808019037690145e-06,
+      "loss": 0.8718,
+      "step": 6371
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.877841478310639e-06,
+      "loss": 0.9016,
+      "step": 6372
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.874881356004864e-06,
+      "loss": 0.8408,
+      "step": 6373
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.871921537139117e-06,
+      "loss": 0.8035,
+      "step": 6374
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8689620220007804e-06,
+      "loss": 0.8481,
+      "step": 6375
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.866002810877224e-06,
+      "loss": 0.7656,
+      "step": 6376
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.863043904055773e-06,
+      "loss": 0.8925,
+      "step": 6377
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.860085301823729e-06,
+      "loss": 0.9094,
+      "step": 6378
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.857127004468367e-06,
+      "loss": 0.7934,
+      "step": 6379
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.854169012276923e-06,
+      "loss": 0.7554,
+      "step": 6380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.851211325536613e-06,
+      "loss": 0.9006,
+      "step": 6381
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.848253944534622e-06,
+      "loss": 0.874,
+      "step": 6382
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8452968695581e-06,
+      "loss": 0.8772,
+      "step": 6383
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.84234010089417e-06,
+      "loss": 0.7829,
+      "step": 6384
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.839383638829931e-06,
+      "loss": 0.9148,
+      "step": 6385
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.836427483652436e-06,
+      "loss": 0.935,
+      "step": 6386
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.833471635648732e-06,
+      "loss": 0.7646,
+      "step": 6387
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.830516095105817e-06,
+      "loss": 0.799,
+      "step": 6388
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.827560862310665e-06,
+      "loss": 0.8195,
+      "step": 6389
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.824605937550224e-06,
+      "loss": 0.8885,
+      "step": 6390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8216513211114045e-06,
+      "loss": 0.8736,
+      "step": 6391
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.818697013281093e-06,
+      "loss": 0.8498,
+      "step": 6392
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.815743014346147e-06,
+      "loss": 0.9157,
+      "step": 6393
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8127893245933864e-06,
+      "loss": 0.8669,
+      "step": 6394
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.809835944309609e-06,
+      "loss": 0.821,
+      "step": 6395
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.806882873781579e-06,
+      "loss": 0.7734,
+      "step": 6396
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.803930113296031e-06,
+      "loss": 0.8597,
+      "step": 6397
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.800977663139666e-06,
+      "loss": 0.856,
+      "step": 6398
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.798025523599165e-06,
+      "loss": 0.8882,
+      "step": 6399
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.795073694961171e-06,
+      "loss": 0.8876,
+      "step": 6400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.792122177512294e-06,
+      "loss": 0.8957,
+      "step": 6401
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.789170971539119e-06,
+      "loss": 0.8418,
+      "step": 6402
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7862200773282035e-06,
+      "loss": 0.8661,
+      "step": 6403
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.783269495166066e-06,
+      "loss": 0.8422,
+      "step": 6404
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.780319225339201e-06,
+      "loss": 0.954,
+      "step": 6405
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.777369268134076e-06,
+      "loss": 0.8806,
+      "step": 6406
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.774419623837114e-06,
+      "loss": 0.8774,
+      "step": 6407
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.771470292734723e-06,
+      "loss": 0.8226,
+      "step": 6408
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.768521275113275e-06,
+      "loss": 0.9495,
+      "step": 6409
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7655725712591055e-06,
+      "loss": 0.8715,
+      "step": 6410
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.762624181458534e-06,
+      "loss": 0.852,
+      "step": 6411
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.759676105997834e-06,
+      "loss": 0.8645,
+      "step": 6412
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7567283451632595e-06,
+      "loss": 0.7627,
+      "step": 6413
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.753780899241027e-06,
+      "loss": 0.8638,
+      "step": 6414
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.750833768517324e-06,
+      "loss": 0.8623,
+      "step": 6415
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.747886953278311e-06,
+      "loss": 0.8555,
+      "step": 6416
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.744940453810113e-06,
+      "loss": 0.8901,
+      "step": 6417
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.741994270398826e-06,
+      "loss": 0.8782,
+      "step": 6418
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.739048403330521e-06,
+      "loss": 0.8884,
+      "step": 6419
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.736102852891227e-06,
+      "loss": 0.8717,
+      "step": 6420
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.73315761936695e-06,
+      "loss": 0.9398,
+      "step": 6421
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.730212703043666e-06,
+      "loss": 0.8559,
+      "step": 6422
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7272681042073175e-06,
+      "loss": 0.777,
+      "step": 6423
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7243238231438176e-06,
+      "loss": 0.923,
+      "step": 6424
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.721379860139044e-06,
+      "loss": 0.874,
+      "step": 6425
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.718436215478849e-06,
+      "loss": 0.8867,
+      "step": 6426
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.715492889449053e-06,
+      "loss": 0.8,
+      "step": 6427
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.712549882335442e-06,
+      "loss": 0.9151,
+      "step": 6428
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.709607194423775e-06,
+      "loss": 0.9185,
+      "step": 6429
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.70666482599978e-06,
+      "loss": 0.8271,
+      "step": 6430
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7037227773491485e-06,
+      "loss": 0.8804,
+      "step": 6431
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7007810487575475e-06,
+      "loss": 0.8627,
+      "step": 6432
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.697839640510609e-06,
+      "loss": 0.9057,
+      "step": 6433
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.694898552893941e-06,
+      "loss": 0.8593,
+      "step": 6434
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.691957786193106e-06,
+      "loss": 0.8185,
+      "step": 6435
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6890173406936485e-06,
+      "loss": 0.8751,
+      "step": 6436
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6860772166810795e-06,
+      "loss": 0.886,
+      "step": 6437
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.683137414440872e-06,
+      "loss": 0.7905,
+      "step": 6438
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.680197934258474e-06,
+      "loss": 0.8253,
+      "step": 6439
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.677258776419304e-06,
+      "loss": 0.8145,
+      "step": 6440
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.67431994120874e-06,
+      "loss": 0.8614,
+      "step": 6441
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.671381428912138e-06,
+      "loss": 0.9394,
+      "step": 6442
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.668443239814819e-06,
+      "loss": 0.894,
+      "step": 6443
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.66550537420207e-06,
+      "loss": 0.8842,
+      "step": 6444
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.662567832359149e-06,
+      "loss": 0.8523,
+      "step": 6445
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.659630614571287e-06,
+      "loss": 0.9265,
+      "step": 6446
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.656693721123678e-06,
+      "loss": 0.8099,
+      "step": 6447
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.653757152301488e-06,
+      "loss": 0.9041,
+      "step": 6448
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6508209083898435e-06,
+      "loss": 0.8663,
+      "step": 6449
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.647884989673849e-06,
+      "loss": 0.8527,
+      "step": 6450
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.644949396438575e-06,
+      "loss": 0.8871,
+      "step": 6451
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.642014128969055e-06,
+      "loss": 0.8163,
+      "step": 6452
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.639079187550299e-06,
+      "loss": 0.8832,
+      "step": 6453
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.63614457246728e-06,
+      "loss": 0.7683,
+      "step": 6454
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.633210284004941e-06,
+      "loss": 0.8882,
+      "step": 6455
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.630276322448188e-06,
+      "loss": 0.9328,
+      "step": 6456
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.627342688081908e-06,
+      "loss": 0.8523,
+      "step": 6457
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.624409381190946e-06,
+      "loss": 0.8733,
+      "step": 6458
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.621476402060116e-06,
+      "loss": 0.828,
+      "step": 6459
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.618543750974202e-06,
+      "loss": 0.8529,
+      "step": 6460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.615611428217959e-06,
+      "loss": 0.9545,
+      "step": 6461
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6126794340761025e-06,
+      "loss": 0.881,
+      "step": 6462
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.609747768833324e-06,
+      "loss": 0.8006,
+      "step": 6463
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.606816432774279e-06,
+      "loss": 0.9042,
+      "step": 6464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.603885426183589e-06,
+      "loss": 0.8887,
+      "step": 6465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.600954749345851e-06,
+      "loss": 0.8777,
+      "step": 6466
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.598024402545624e-06,
+      "loss": 0.8134,
+      "step": 6467
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.595094386067428e-06,
+      "loss": 0.8126,
+      "step": 6468
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.592164700195775e-06,
+      "loss": 0.8859,
+      "step": 6469
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.589235345215117e-06,
+      "loss": 0.8383,
+      "step": 6470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.58630632140989e-06,
+      "loss": 0.8804,
+      "step": 6471
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.583377629064494e-06,
+      "loss": 0.873,
+      "step": 6472
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.580449268463296e-06,
+      "loss": 0.8405,
+      "step": 6473
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5775212398906295e-06,
+      "loss": 0.8979,
+      "step": 6474
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.574593543630803e-06,
+      "loss": 0.8631,
+      "step": 6475
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.571666179968079e-06,
+      "loss": 0.8226,
+      "step": 6476
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.568739149186705e-06,
+      "loss": 0.9005,
+      "step": 6477
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.565812451570881e-06,
+      "loss": 0.8969,
+      "step": 6478
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.562886087404783e-06,
+      "loss": 0.8566,
+      "step": 6479
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5599600569725495e-06,
+      "loss": 0.8665,
+      "step": 6480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.557034360558294e-06,
+      "loss": 0.8398,
+      "step": 6481
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.554108998446096e-06,
+      "loss": 0.8789,
+      "step": 6482
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.551183970919993e-06,
+      "loss": 0.868,
+      "step": 6483
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.548259278263999e-06,
+      "loss": 0.8816,
+      "step": 6484
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.545334920762096e-06,
+      "loss": 0.9461,
+      "step": 6485
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.542410898698226e-06,
+      "loss": 0.861,
+      "step": 6486
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.539487212356306e-06,
+      "loss": 0.9435,
+      "step": 6487
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.536563862020218e-06,
+      "loss": 0.7966,
+      "step": 6488
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.533640847973809e-06,
+      "loss": 0.8929,
+      "step": 6489
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.530718170500896e-06,
+      "loss": 0.8604,
+      "step": 6490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.527795829885259e-06,
+      "loss": 0.7896,
+      "step": 6491
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.524873826410658e-06,
+      "loss": 0.8772,
+      "step": 6492
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.521952160360803e-06,
+      "loss": 0.8617,
+      "step": 6493
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.519030832019383e-06,
+      "loss": 0.8434,
+      "step": 6494
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.516109841670051e-06,
+      "loss": 0.8626,
+      "step": 6495
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.513189189596422e-06,
+      "loss": 0.8754,
+      "step": 6496
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.510268876082089e-06,
+      "loss": 0.84,
+      "step": 6497
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.507348901410604e-06,
+      "loss": 0.8539,
+      "step": 6498
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.504429265865486e-06,
+      "loss": 0.8328,
+      "step": 6499
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.501509969730224e-06,
+      "loss": 0.936,
+      "step": 6500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.498591013288276e-06,
+      "loss": 0.9598,
+      "step": 6501
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.495672396823061e-06,
+      "loss": 0.7855,
+      "step": 6502
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.492754120617966e-06,
+      "loss": 0.841,
+      "step": 6503
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.489836184956353e-06,
+      "loss": 0.753,
+      "step": 6504
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.486918590121544e-06,
+      "loss": 0.8749,
+      "step": 6505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.484001336396828e-06,
+      "loss": 0.8985,
+      "step": 6506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.481084424065461e-06,
+      "loss": 0.8588,
+      "step": 6507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.478167853410668e-06,
+      "loss": 0.8661,
+      "step": 6508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.475251624715641e-06,
+      "loss": 0.8096,
+      "step": 6509
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.472335738263534e-06,
+      "loss": 0.8288,
+      "step": 6510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.46942019433747e-06,
+      "loss": 0.8443,
+      "step": 6511
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.466504993220548e-06,
+      "loss": 0.7979,
+      "step": 6512
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.463590135195816e-06,
+      "loss": 0.8701,
+      "step": 6513
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.460675620546305e-06,
+      "loss": 0.8994,
+      "step": 6514
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.457761449555e-06,
+      "loss": 0.8469,
+      "step": 6515
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.454847622504867e-06,
+      "loss": 0.8536,
+      "step": 6516
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4519341396788235e-06,
+      "loss": 0.8512,
+      "step": 6517
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4490210013597635e-06,
+      "loss": 0.9255,
+      "step": 6518
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.446108207830543e-06,
+      "loss": 0.8811,
+      "step": 6519
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4431957593739845e-06,
+      "loss": 0.8556,
+      "step": 6520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.440283656272881e-06,
+      "loss": 0.9033,
+      "step": 6521
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4373718988099896e-06,
+      "loss": 0.8801,
+      "step": 6522
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4344604872680295e-06,
+      "loss": 0.7749,
+      "step": 6523
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.431549421929694e-06,
+      "loss": 0.8296,
+      "step": 6524
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.42863870307764e-06,
+      "loss": 0.893,
+      "step": 6525
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4257283309944804e-06,
+      "loss": 0.9114,
+      "step": 6526
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.42281830596282e-06,
+      "loss": 0.8221,
+      "step": 6527
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.419908628265203e-06,
+      "loss": 0.9089,
+      "step": 6528
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.416999298184153e-06,
+      "loss": 0.8927,
+      "step": 6529
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.414090316002161e-06,
+      "loss": 0.919,
+      "step": 6530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4111816820016745e-06,
+      "loss": 0.8461,
+      "step": 6531
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4082733964651166e-06,
+      "loss": 0.8788,
+      "step": 6532
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.405365459674876e-06,
+      "loss": 0.7693,
+      "step": 6533
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4024578719133e-06,
+      "loss": 0.8664,
+      "step": 6534
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.39955063346271e-06,
+      "loss": 0.8864,
+      "step": 6535
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.396643744605391e-06,
+      "loss": 0.8664,
+      "step": 6536
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.393737205623591e-06,
+      "loss": 0.9374,
+      "step": 6537
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.390831016799527e-06,
+      "loss": 0.7944,
+      "step": 6538
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.387925178415384e-06,
+      "loss": 0.7898,
+      "step": 6539
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.385019690753311e-06,
+      "loss": 0.8849,
+      "step": 6540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.38211455409542e-06,
+      "loss": 0.8924,
+      "step": 6541
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.379209768723791e-06,
+      "loss": 0.8611,
+      "step": 6542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.376305334920473e-06,
+      "loss": 0.919,
+      "step": 6543
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.373401252967475e-06,
+      "loss": 0.8767,
+      "step": 6544
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.370497523146777e-06,
+      "loss": 0.8699,
+      "step": 6545
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.367594145740324e-06,
+      "loss": 0.8723,
+      "step": 6546
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.364691121030021e-06,
+      "loss": 0.8594,
+      "step": 6547
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.361788449297748e-06,
+      "loss": 0.8647,
+      "step": 6548
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.358886130825344e-06,
+      "loss": 0.8868,
+      "step": 6549
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.355984165894613e-06,
+      "loss": 0.8622,
+      "step": 6550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.353082554787335e-06,
+      "loss": 0.8282,
+      "step": 6551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.350181297785242e-06,
+      "loss": 0.8448,
+      "step": 6552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.347280395170041e-06,
+      "loss": 0.887,
+      "step": 6553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.344379847223398e-06,
+      "loss": 0.8497,
+      "step": 6554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.34147965422695e-06,
+      "loss": 0.8387,
+      "step": 6555
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.338579816462298e-06,
+      "loss": 0.8356,
+      "step": 6556
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.335680334211005e-06,
+      "loss": 0.8928,
+      "step": 6557
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.332781207754605e-06,
+      "loss": 0.8872,
+      "step": 6558
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.329882437374595e-06,
+      "loss": 0.8625,
+      "step": 6559
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.326984023352435e-06,
+      "loss": 0.9399,
+      "step": 6560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.324085965969551e-06,
+      "loss": 0.8805,
+      "step": 6561
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.321188265507342e-06,
+      "loss": 0.8691,
+      "step": 6562
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.318290922247163e-06,
+      "loss": 0.8522,
+      "step": 6563
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.31539393647034e-06,
+      "loss": 0.8776,
+      "step": 6564
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3124973084581585e-06,
+      "loss": 0.9063,
+      "step": 6565
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.309601038491874e-06,
+      "loss": 0.8499,
+      "step": 6566
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.306705126852708e-06,
+      "loss": 0.7966,
+      "step": 6567
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.303809573821842e-06,
+      "loss": 0.8888,
+      "step": 6568
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.300914379680427e-06,
+      "loss": 0.893,
+      "step": 6569
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.298019544709579e-06,
+      "loss": 0.8526,
+      "step": 6570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.295125069190378e-06,
+      "loss": 0.8289,
+      "step": 6571
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.292230953403866e-06,
+      "loss": 0.8787,
+      "step": 6572
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.289337197631056e-06,
+      "loss": 0.9244,
+      "step": 6573
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.286443802152926e-06,
+      "loss": 0.854,
+      "step": 6574
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.283550767250413e-06,
+      "loss": 0.8612,
+      "step": 6575
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.280658093204422e-06,
+      "loss": 0.9094,
+      "step": 6576
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.277765780295828e-06,
+      "loss": 0.9195,
+      "step": 6577
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.274873828805459e-06,
+      "loss": 0.8583,
+      "step": 6578
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.271982239014121e-06,
+      "loss": 0.9044,
+      "step": 6579
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.269091011202576e-06,
+      "loss": 0.8744,
+      "step": 6580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.266200145651554e-06,
+      "loss": 0.8373,
+      "step": 6581
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.263309642641751e-06,
+      "loss": 0.7882,
+      "step": 6582
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.260419502453828e-06,
+      "loss": 0.8191,
+      "step": 6583
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.257529725368405e-06,
+      "loss": 0.9443,
+      "step": 6584
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.254640311666072e-06,
+      "loss": 0.8158,
+      "step": 6585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.251751261627386e-06,
+      "loss": 0.7915,
+      "step": 6586
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2488625755328615e-06,
+      "loss": 0.903,
+      "step": 6587
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.245974253662988e-06,
+      "loss": 0.9158,
+      "step": 6588
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.243086296298206e-06,
+      "loss": 0.8954,
+      "step": 6589
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.240198703718932e-06,
+      "loss": 0.851,
+      "step": 6590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.237311476205541e-06,
+      "loss": 0.8721,
+      "step": 6591
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.234424614038375e-06,
+      "loss": 0.9142,
+      "step": 6592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.23153811749774e-06,
+      "loss": 0.8384,
+      "step": 6593
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2286519868639095e-06,
+      "loss": 0.8855,
+      "step": 6594
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.225766222417112e-06,
+      "loss": 0.8593,
+      "step": 6595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.222880824437549e-06,
+      "loss": 0.8381,
+      "step": 6596
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.219995793205389e-06,
+      "loss": 0.9166,
+      "step": 6597
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.217111129000759e-06,
+      "loss": 0.8383,
+      "step": 6598
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.214226832103747e-06,
+      "loss": 0.7724,
+      "step": 6599
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.211342902794413e-06,
+      "loss": 0.9001,
+      "step": 6600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.20845934135278e-06,
+      "loss": 0.8594,
+      "step": 6601
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.205576148058828e-06,
+      "loss": 0.8248,
+      "step": 6602
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.202693323192512e-06,
+      "loss": 0.9044,
+      "step": 6603
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.199810867033745e-06,
+      "loss": 0.8227,
+      "step": 6604
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1969287798624034e-06,
+      "loss": 0.8708,
+      "step": 6605
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.19404706195833e-06,
+      "loss": 0.847,
+      "step": 6606
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.191165713601334e-06,
+      "loss": 0.8006,
+      "step": 6607
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.188284735071177e-06,
+      "loss": 0.8964,
+      "step": 6608
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1854041266476075e-06,
+      "loss": 0.7924,
+      "step": 6609
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.182523888610316e-06,
+      "loss": 0.9014,
+      "step": 6610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.179644021238966e-06,
+      "loss": 0.8911,
+      "step": 6611
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.176764524813187e-06,
+      "loss": 0.8349,
+      "step": 6612
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.173885399612566e-06,
+      "loss": 0.862,
+      "step": 6613
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.171006645916662e-06,
+      "loss": 0.9227,
+      "step": 6614
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1681282640049885e-06,
+      "loss": 0.8785,
+      "step": 6615
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.165250254157032e-06,
+      "loss": 0.8528,
+      "step": 6616
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.162372616652239e-06,
+      "loss": 0.9465,
+      "step": 6617
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.159495351770017e-06,
+      "loss": 0.84,
+      "step": 6618
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.156618459789742e-06,
+      "loss": 0.8854,
+      "step": 6619
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.153741940990749e-06,
+      "loss": 0.8318,
+      "step": 6620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.150865795652345e-06,
+      "loss": 0.9472,
+      "step": 6621
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1479900240537956e-06,
+      "loss": 0.8433,
+      "step": 6622
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.145114626474324e-06,
+      "loss": 0.8867,
+      "step": 6623
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.142239603193128e-06,
+      "loss": 0.8636,
+      "step": 6624
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.139364954489363e-06,
+      "loss": 0.8384,
+      "step": 6625
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.136490680642146e-06,
+      "loss": 0.8149,
+      "step": 6626
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.133616781930565e-06,
+      "loss": 0.8616,
+      "step": 6627
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.130743258633667e-06,
+      "loss": 0.8647,
+      "step": 6628
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.127870111030458e-06,
+      "loss": 0.8261,
+      "step": 6629
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.124997339399916e-06,
+      "loss": 0.8658,
+      "step": 6630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.122124944020978e-06,
+      "loss": 0.8186,
+      "step": 6631
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.119252925172549e-06,
+      "loss": 0.8949,
+      "step": 6632
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.116381283133489e-06,
+      "loss": 0.8821,
+      "step": 6633
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.113510018182628e-06,
+      "loss": 0.8231,
+      "step": 6634
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.110639130598759e-06,
+      "loss": 0.8716,
+      "step": 6635
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.107768620660633e-06,
+      "loss": 0.9398,
+      "step": 6636
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.104898488646971e-06,
+      "loss": 0.8513,
+      "step": 6637
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.102028734836456e-06,
+      "loss": 0.8829,
+      "step": 6638
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.099159359507729e-06,
+      "loss": 0.8352,
+      "step": 6639
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0962903629394e-06,
+      "loss": 0.9062,
+      "step": 6640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.093421745410042e-06,
+      "loss": 0.8672,
+      "step": 6641
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.090553507198187e-06,
+      "loss": 0.9243,
+      "step": 6642
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.08768564858233e-06,
+      "loss": 0.9099,
+      "step": 6643
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0848181698409384e-06,
+      "loss": 0.8976,
+      "step": 6644
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.081951071252432e-06,
+      "loss": 0.8528,
+      "step": 6645
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.079084353095202e-06,
+      "loss": 0.9259,
+      "step": 6646
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0762180156475924e-06,
+      "loss": 0.8725,
+      "step": 6647
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.07335205918792e-06,
+      "loss": 0.8155,
+      "step": 6648
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.070486483994462e-06,
+      "loss": 0.9251,
+      "step": 6649
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.067621290345455e-06,
+      "loss": 0.8762,
+      "step": 6650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0647564785191006e-06,
+      "loss": 0.8348,
+      "step": 6651
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.061892048793568e-06,
+      "loss": 0.8909,
+      "step": 6652
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.059028001446979e-06,
+      "loss": 0.8211,
+      "step": 6653
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.056164336757426e-06,
+      "loss": 0.8503,
+      "step": 6654
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.053301055002964e-06,
+      "loss": 0.8696,
+      "step": 6655
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.050438156461613e-06,
+      "loss": 0.8639,
+      "step": 6656
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.047575641411345e-06,
+      "loss": 0.9043,
+      "step": 6657
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.044713510130108e-06,
+      "loss": 0.8578,
+      "step": 6658
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.041851762895805e-06,
+      "loss": 0.9062,
+      "step": 6659
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.038990399986302e-06,
+      "loss": 0.8625,
+      "step": 6660
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0361294216794284e-06,
+      "loss": 0.8695,
+      "step": 6661
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.03326882825298e-06,
+      "loss": 0.865,
+      "step": 6662
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0304086199847105e-06,
+      "loss": 0.9014,
+      "step": 6663
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.027548797152336e-06,
+      "loss": 0.8232,
+      "step": 6664
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.024689360033542e-06,
+      "loss": 0.9288,
+      "step": 6665
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.021830308905963e-06,
+      "loss": 0.8176,
+      "step": 6666
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.018971644047216e-06,
+      "loss": 0.8716,
+      "step": 6667
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.016113365734861e-06,
+      "loss": 0.8491,
+      "step": 6668
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.013255474246431e-06,
+      "loss": 0.8919,
+      "step": 6669
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0103979698594215e-06,
+      "loss": 0.7839,
+      "step": 6670
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.007540852851283e-06,
+      "loss": 0.9113,
+      "step": 6671
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.004684123499436e-06,
+      "loss": 0.9276,
+      "step": 6672
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0018277820812636e-06,
+      "loss": 0.8448,
+      "step": 6673
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.998971828874102e-06,
+      "loss": 0.8665,
+      "step": 6674
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.996116264155261e-06,
+      "loss": 0.8063,
+      "step": 6675
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.993261088202005e-06,
+      "loss": 0.855,
+      "step": 6676
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.990406301291563e-06,
+      "loss": 0.9194,
+      "step": 6677
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.987551903701128e-06,
+      "loss": 0.853,
+      "step": 6678
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.984697895707854e-06,
+      "loss": 0.8688,
+      "step": 6679
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9818442775888595e-06,
+      "loss": 0.8587,
+      "step": 6680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.978991049621216e-06,
+      "loss": 0.8901,
+      "step": 6681
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.97613821208197e-06,
+      "loss": 0.8992,
+      "step": 6682
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.973285765248122e-06,
+      "loss": 0.8838,
+      "step": 6683
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.970433709396635e-06,
+      "loss": 0.9127,
+      "step": 6684
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9675820448044364e-06,
+      "loss": 0.7927,
+      "step": 6685
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.964730771748415e-06,
+      "loss": 0.9162,
+      "step": 6686
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.961879890505419e-06,
+      "loss": 0.8442,
+      "step": 6687
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.959029401352262e-06,
+      "loss": 0.9325,
+      "step": 6688
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.956179304565722e-06,
+      "loss": 0.7886,
+      "step": 6689
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.953329600422524e-06,
+      "loss": 0.8289,
+      "step": 6690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9504802891993805e-06,
+      "loss": 0.8508,
+      "step": 6691
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.947631371172943e-06,
+      "loss": 0.8489,
+      "step": 6692
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.944782846619837e-06,
+      "loss": 0.8293,
+      "step": 6693
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.941934715816642e-06,
+      "loss": 0.9026,
+      "step": 6694
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.939086979039905e-06,
+      "loss": 0.8591,
+      "step": 6695
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.936239636566137e-06,
+      "loss": 0.8726,
+      "step": 6696
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9333926886718e-06,
+      "loss": 0.7912,
+      "step": 6697
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.930546135633327e-06,
+      "loss": 0.8203,
+      "step": 6698
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.927699977727113e-06,
+      "loss": 0.8656,
+      "step": 6699
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.924854215229509e-06,
+      "loss": 0.9431,
+      "step": 6700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.922008848416827e-06,
+      "loss": 0.8725,
+      "step": 6701
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.919163877565351e-06,
+      "loss": 0.8477,
+      "step": 6702
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.916319302951316e-06,
+      "loss": 0.7881,
+      "step": 6703
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9134751248509236e-06,
+      "loss": 0.8477,
+      "step": 6704
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.910631343540332e-06,
+      "loss": 0.8401,
+      "step": 6705
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9077879592956675e-06,
+      "loss": 0.8634,
+      "step": 6706
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.904944972393014e-06,
+      "loss": 0.8489,
+      "step": 6707
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.902102383108415e-06,
+      "loss": 0.8371,
+      "step": 6708
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.899260191717878e-06,
+      "loss": 0.8784,
+      "step": 6709
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.896418398497377e-06,
+      "loss": 0.8365,
+      "step": 6710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.893577003722834e-06,
+      "loss": 0.8215,
+      "step": 6711
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.890736007670144e-06,
+      "loss": 0.8566,
+      "step": 6712
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.887895410615159e-06,
+      "loss": 0.8722,
+      "step": 6713
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.885055212833696e-06,
+      "loss": 0.856,
+      "step": 6714
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.882215414601527e-06,
+      "loss": 0.828,
+      "step": 6715
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.879376016194387e-06,
+      "loss": 0.9222,
+      "step": 6716
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.876537017887978e-06,
+      "loss": 0.8885,
+      "step": 6717
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.873698419957952e-06,
+      "loss": 0.8368,
+      "step": 6718
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8708602226799336e-06,
+      "loss": 0.9246,
+      "step": 6719
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8680224263295045e-06,
+      "loss": 0.8242,
+      "step": 6720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8651850311822025e-06,
+      "loss": 0.851,
+      "step": 6721
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.862348037513533e-06,
+      "loss": 0.7433,
+      "step": 6722
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.85951144559896e-06,
+      "loss": 0.9115,
+      "step": 6723
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.856675255713905e-06,
+      "loss": 0.8445,
+      "step": 6724
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.853839468133755e-06,
+      "loss": 0.8051,
+      "step": 6725
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.851004083133862e-06,
+      "loss": 0.8776,
+      "step": 6726
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.848169100989529e-06,
+      "loss": 0.8905,
+      "step": 6727
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8453345219760275e-06,
+      "loss": 0.8548,
+      "step": 6728
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.842500346368583e-06,
+      "loss": 0.8943,
+      "step": 6729
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.839666574442389e-06,
+      "loss": 0.9197,
+      "step": 6730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.836833206472596e-06,
+      "loss": 0.9423,
+      "step": 6731
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.834000242734317e-06,
+      "loss": 0.8779,
+      "step": 6732
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8311676835026174e-06,
+      "loss": 0.8434,
+      "step": 6733
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.828335529052541e-06,
+      "loss": 0.819,
+      "step": 6734
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.825503779659078e-06,
+      "loss": 0.837,
+      "step": 6735
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.822672435597172e-06,
+      "loss": 0.8924,
+      "step": 6736
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.819841497141758e-06,
+      "loss": 0.8697,
+      "step": 6737
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.817010964567702e-06,
+      "loss": 0.8769,
+      "step": 6738
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.814180838149838e-06,
+      "loss": 0.8743,
+      "step": 6739
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.811351118162969e-06,
+      "loss": 0.9733,
+      "step": 6740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8085218048818505e-06,
+      "loss": 0.8451,
+      "step": 6741
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.805692898581196e-06,
+      "loss": 0.8221,
+      "step": 6742
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8028643995356925e-06,
+      "loss": 0.8516,
+      "step": 6743
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.800036308019974e-06,
+      "loss": 0.8651,
+      "step": 6744
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.797208624308641e-06,
+      "loss": 1.0115,
+      "step": 6745
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.79438134867625e-06,
+      "loss": 0.8865,
+      "step": 6746
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.791554481397328e-06,
+      "loss": 0.8298,
+      "step": 6747
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.788728022746348e-06,
+      "loss": 0.775,
+      "step": 6748
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7859019729977564e-06,
+      "loss": 0.8182,
+      "step": 6749
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.783076332425957e-06,
+      "loss": 0.8049,
+      "step": 6750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.780251101305308e-06,
+      "loss": 0.8333,
+      "step": 6751
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.777426279910125e-06,
+      "loss": 0.8715,
+      "step": 6752
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.774601868514702e-06,
+      "loss": 0.8992,
+      "step": 6753
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.771777867393275e-06,
+      "loss": 0.7885,
+      "step": 6754
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.768954276820042e-06,
+      "loss": 0.838,
+      "step": 6755
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.766131097069174e-06,
+      "loss": 0.7929,
+      "step": 6756
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7633083284147904e-06,
+      "loss": 0.8871,
+      "step": 6757
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.760485971130969e-06,
+      "loss": 0.8727,
+      "step": 6758
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7576640254917604e-06,
+      "loss": 0.7898,
+      "step": 6759
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7548424917711596e-06,
+      "loss": 0.8924,
+      "step": 6760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7520213702431395e-06,
+      "loss": 0.8682,
+      "step": 6761
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.749200661181611e-06,
+      "loss": 0.9036,
+      "step": 6762
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.746380364860469e-06,
+      "loss": 0.9268,
+      "step": 6763
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7435604815535475e-06,
+      "loss": 0.9178,
+      "step": 6764
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.740741011534649e-06,
+      "loss": 0.8417,
+      "step": 6765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7379219550775415e-06,
+      "loss": 0.794,
+      "step": 6766
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.735103312455945e-06,
+      "loss": 0.822,
+      "step": 6767
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.732285083943537e-06,
+      "loss": 0.8822,
+      "step": 6768
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.729467269813966e-06,
+      "loss": 0.8917,
+      "step": 6769
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.726649870340833e-06,
+      "loss": 0.8166,
+      "step": 6770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.723832885797691e-06,
+      "loss": 0.8109,
+      "step": 6771
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.721016316458068e-06,
+      "loss": 0.8877,
+      "step": 6772
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.718200162595449e-06,
+      "loss": 0.833,
+      "step": 6773
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.715384424483268e-06,
+      "loss": 0.8858,
+      "step": 6774
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7125691023949224e-06,
+      "loss": 0.8752,
+      "step": 6775
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.709754196603781e-06,
+      "loss": 0.9094,
+      "step": 6776
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.706939707383159e-06,
+      "loss": 0.9098,
+      "step": 6777
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.704125635006329e-06,
+      "loss": 0.8543,
+      "step": 6778
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.701311979746539e-06,
+      "loss": 0.8513,
+      "step": 6779
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6984987418769825e-06,
+      "loss": 0.8388,
+      "step": 6780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.695685921670812e-06,
+      "loss": 0.8683,
+      "step": 6781
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.692873519401154e-06,
+      "loss": 0.8842,
+      "step": 6782
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.690061535341076e-06,
+      "loss": 0.8382,
+      "step": 6783
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6872499697636195e-06,
+      "loss": 0.8182,
+      "step": 6784
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.684438822941775e-06,
+      "loss": 0.8218,
+      "step": 6785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.681628095148502e-06,
+      "loss": 0.891,
+      "step": 6786
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.678817786656711e-06,
+      "loss": 0.9148,
+      "step": 6787
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6760078977392706e-06,
+      "loss": 0.8981,
+      "step": 6788
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.673198428669022e-06,
+      "loss": 0.8765,
+      "step": 6789
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.67038937971875e-06,
+      "loss": 0.8822,
+      "step": 6790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6675807511612055e-06,
+      "loss": 0.8251,
+      "step": 6791
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.664772543269101e-06,
+      "loss": 0.8477,
+      "step": 6792
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.661964756315106e-06,
+      "loss": 0.8858,
+      "step": 6793
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.659157390571842e-06,
+      "loss": 0.8918,
+      "step": 6794
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.656350446311901e-06,
+      "loss": 0.885,
+      "step": 6795
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.653543923807833e-06,
+      "loss": 0.9195,
+      "step": 6796
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.650737823332136e-06,
+      "loss": 0.8423,
+      "step": 6797
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6479321451572785e-06,
+      "loss": 0.8767,
+      "step": 6798
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.645126889555686e-06,
+      "loss": 0.868,
+      "step": 6799
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.642322056799732e-06,
+      "loss": 0.8543,
+      "step": 6800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6395176471617665e-06,
+      "loss": 0.8568,
+      "step": 6801
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.636713660914087e-06,
+      "loss": 0.8207,
+      "step": 6802
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.633910098328947e-06,
+      "loss": 0.8882,
+      "step": 6803
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.631106959678575e-06,
+      "loss": 0.9216,
+      "step": 6804
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6283042452351414e-06,
+      "loss": 0.8764,
+      "step": 6805
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.625501955270777e-06,
+      "loss": 0.9194,
+      "step": 6806
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.622700090057581e-06,
+      "loss": 0.9568,
+      "step": 6807
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.619898649867612e-06,
+      "loss": 0.8529,
+      "step": 6808
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6170976349728754e-06,
+      "loss": 0.8105,
+      "step": 6809
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.614297045645339e-06,
+      "loss": 0.8474,
+      "step": 6810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.611496882156941e-06,
+      "loss": 0.8887,
+      "step": 6811
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6086971447795625e-06,
+      "loss": 0.8322,
+      "step": 6812
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.605897833785048e-06,
+      "loss": 0.8451,
+      "step": 6813
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.603098949445209e-06,
+      "loss": 0.8119,
+      "step": 6814
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.600300492031806e-06,
+      "loss": 0.8524,
+      "step": 6815
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.597502461816557e-06,
+      "loss": 0.8023,
+      "step": 6816
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5947048590711515e-06,
+      "loss": 0.886,
+      "step": 6817
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5919076840672215e-06,
+      "loss": 0.9441,
+      "step": 6818
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.58911093707637e-06,
+      "loss": 0.8374,
+      "step": 6819
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5863146183701454e-06,
+      "loss": 0.9213,
+      "step": 6820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.583518728220071e-06,
+      "loss": 0.8208,
+      "step": 6821
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.580723266897616e-06,
+      "loss": 0.7743,
+      "step": 6822
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.577928234674208e-06,
+      "loss": 0.8125,
+      "step": 6823
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.575133631821243e-06,
+      "loss": 0.8553,
+      "step": 6824
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.572339458610068e-06,
+      "loss": 0.9005,
+      "step": 6825
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5695457153119806e-06,
+      "loss": 0.8515,
+      "step": 6826
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.566752402198256e-06,
+      "loss": 0.8425,
+      "step": 6827
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.563959519540114e-06,
+      "loss": 0.8204,
+      "step": 6828
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.561167067608728e-06,
+      "loss": 0.8813,
+      "step": 6829
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.558375046675244e-06,
+      "loss": 0.8853,
+      "step": 6830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.555583457010763e-06,
+      "loss": 0.8008,
+      "step": 6831
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.552792298886335e-06,
+      "loss": 0.8757,
+      "step": 6832
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.550001572572969e-06,
+      "loss": 0.8345,
+      "step": 6833
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.547211278341646e-06,
+      "loss": 0.8812,
+      "step": 6834
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.544421416463293e-06,
+      "loss": 0.8044,
+      "step": 6835
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.541631987208789e-06,
+      "loss": 0.9097,
+      "step": 6836
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.538842990848992e-06,
+      "loss": 0.9218,
+      "step": 6837
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.536054427654698e-06,
+      "loss": 0.9058,
+      "step": 6838
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.533266297896667e-06,
+      "loss": 0.8325,
+      "step": 6839
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.530478601845624e-06,
+      "loss": 0.8635,
+      "step": 6840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.527691339772241e-06,
+      "loss": 0.9226,
+      "step": 6841
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.52490451194716e-06,
+      "loss": 0.8196,
+      "step": 6842
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.522118118640964e-06,
+      "loss": 0.7982,
+      "step": 6843
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.519332160124215e-06,
+      "loss": 0.8737,
+      "step": 6844
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.516546636667417e-06,
+      "loss": 0.8971,
+      "step": 6845
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.513761548541032e-06,
+      "loss": 0.8631,
+      "step": 6846
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.51097689601549e-06,
+      "loss": 0.9479,
+      "step": 6847
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.508192679361169e-06,
+      "loss": 0.8609,
+      "step": 6848
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.505408898848407e-06,
+      "loss": 0.9259,
+      "step": 6849
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.502625554747508e-06,
+      "loss": 0.8161,
+      "step": 6850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.499842647328722e-06,
+      "loss": 0.821,
+      "step": 6851
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.497060176862259e-06,
+      "loss": 0.898,
+      "step": 6852
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.49427814361829e-06,
+      "loss": 0.7969,
+      "step": 6853
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.491496547866948e-06,
+      "loss": 0.8015,
+      "step": 6854
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.488715389878311e-06,
+      "loss": 0.8685,
+      "step": 6855
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.485934669922428e-06,
+      "loss": 0.8953,
+      "step": 6856
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.483154388269295e-06,
+      "loss": 0.8685,
+      "step": 6857
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.480374545188866e-06,
+      "loss": 0.8713,
+      "step": 6858
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4775951409510634e-06,
+      "loss": 0.9215,
+      "step": 6859
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.474816175825754e-06,
+      "loss": 0.8287,
+      "step": 6860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.472037650082766e-06,
+      "loss": 0.8194,
+      "step": 6861
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.469259563991894e-06,
+      "loss": 0.8366,
+      "step": 6862
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.466481917822874e-06,
+      "loss": 0.8207,
+      "step": 6863
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.46370471184541e-06,
+      "loss": 0.8881,
+      "step": 6864
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.46092794632916e-06,
+      "loss": 0.9004,
+      "step": 6865
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.458151621543744e-06,
+      "loss": 0.8388,
+      "step": 6866
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.455375737758732e-06,
+      "loss": 0.8589,
+      "step": 6867
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.452600295243653e-06,
+      "loss": 0.848,
+      "step": 6868
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.449825294267998e-06,
+      "loss": 0.8713,
+      "step": 6869
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4470507351012116e-06,
+      "loss": 0.8757,
+      "step": 6870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.444276618012688e-06,
+      "loss": 0.9083,
+      "step": 6871
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.441502943271797e-06,
+      "loss": 0.8393,
+      "step": 6872
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.438729711147848e-06,
+      "loss": 0.7921,
+      "step": 6873
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4359569219101115e-06,
+      "loss": 0.8044,
+      "step": 6874
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.433184575827824e-06,
+      "loss": 0.7937,
+      "step": 6875
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.430412673170167e-06,
+      "loss": 0.829,
+      "step": 6876
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.427641214206288e-06,
+      "loss": 0.9087,
+      "step": 6877
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.424870199205283e-06,
+      "loss": 0.8221,
+      "step": 6878
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.422099628436217e-06,
+      "loss": 0.8393,
+      "step": 6879
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4193295021681e-06,
+      "loss": 0.9706,
+      "step": 6880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4165598206699e-06,
+      "loss": 0.7783,
+      "step": 6881
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.413790584210551e-06,
+      "loss": 0.8746,
+      "step": 6882
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.411021793058936e-06,
+      "loss": 0.831,
+      "step": 6883
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.408253447483892e-06,
+      "loss": 0.8771,
+      "step": 6884
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.405485547754226e-06,
+      "loss": 0.8944,
+      "step": 6885
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.402718094138688e-06,
+      "loss": 0.8667,
+      "step": 6886
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.399951086905985e-06,
+      "loss": 0.8832,
+      "step": 6887
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.397184526324792e-06,
+      "loss": 0.8276,
+      "step": 6888
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.394418412663737e-06,
+      "loss": 0.8972,
+      "step": 6889
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.391652746191398e-06,
+      "loss": 0.9633,
+      "step": 6890
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.388887527176308e-06,
+      "loss": 0.8875,
+      "step": 6891
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.38612275588697e-06,
+      "loss": 0.9524,
+      "step": 6892
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.383358432591832e-06,
+      "loss": 0.9006,
+      "step": 6893
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.380594557559298e-06,
+      "loss": 0.8932,
+      "step": 6894
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3778311310577405e-06,
+      "loss": 0.8781,
+      "step": 6895
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.375068153355474e-06,
+      "loss": 0.8193,
+      "step": 6896
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.372305624720775e-06,
+      "loss": 0.8666,
+      "step": 6897
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.369543545421883e-06,
+      "loss": 0.8837,
+      "step": 6898
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.366781915726985e-06,
+      "loss": 0.9024,
+      "step": 6899
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.364020735904223e-06,
+      "loss": 0.7815,
+      "step": 6900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.361260006221702e-06,
+      "loss": 0.8506,
+      "step": 6901
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.358499726947488e-06,
+      "loss": 0.7472,
+      "step": 6902
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.355739898349589e-06,
+      "loss": 0.9188,
+      "step": 6903
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.352980520695974e-06,
+      "loss": 0.9032,
+      "step": 6904
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3502215942545785e-06,
+      "loss": 0.8187,
+      "step": 6905
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.347463119293283e-06,
+      "loss": 0.8073,
+      "step": 6906
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.344705096079923e-06,
+      "loss": 0.894,
+      "step": 6907
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.341947524882301e-06,
+      "loss": 0.9164,
+      "step": 6908
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.339190405968169e-06,
+      "loss": 0.8953,
+      "step": 6909
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.336433739605227e-06,
+      "loss": 0.8347,
+      "step": 6910
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.333677526061145e-06,
+      "loss": 0.931,
+      "step": 6911
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.330921765603549e-06,
+      "loss": 0.8895,
+      "step": 6912
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.328166458500006e-06,
+      "loss": 0.9088,
+      "step": 6913
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.325411605018056e-06,
+      "loss": 0.9155,
+      "step": 6914
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.322657205425184e-06,
+      "loss": 0.7743,
+      "step": 6915
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.31990325998883e-06,
+      "loss": 0.9333,
+      "step": 6916
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3171497689764016e-06,
+      "loss": 0.8138,
+      "step": 6917
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.314396732655253e-06,
+      "loss": 0.8464,
+      "step": 6918
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3116441512926895e-06,
+      "loss": 0.9124,
+      "step": 6919
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.308892025155989e-06,
+      "loss": 0.9912,
+      "step": 6920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.306140354512369e-06,
+      "loss": 0.8272,
+      "step": 6921
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.303389139629007e-06,
+      "loss": 0.8288,
+      "step": 6922
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.30063838077304e-06,
+      "loss": 0.8583,
+      "step": 6923
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.297888078211564e-06,
+      "loss": 0.8697,
+      "step": 6924
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.29513823221162e-06,
+      "loss": 0.9134,
+      "step": 6925
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2923888430402085e-06,
+      "loss": 0.838,
+      "step": 6926
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.289639910964292e-06,
+      "loss": 0.8447,
+      "step": 6927
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.286891436250785e-06,
+      "loss": 0.8054,
+      "step": 6928
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.284143419166548e-06,
+      "loss": 0.9314,
+      "step": 6929
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.281395859978414e-06,
+      "loss": 0.8573,
+      "step": 6930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.278648758953161e-06,
+      "loss": 0.8155,
+      "step": 6931
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2759021163575184e-06,
+      "loss": 0.843,
+      "step": 6932
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2731559324581885e-06,
+      "loss": 0.9121,
+      "step": 6933
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.27041020752181e-06,
+      "loss": 0.8975,
+      "step": 6934
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.267664941814984e-06,
+      "loss": 0.8455,
+      "step": 6935
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.26492013560427e-06,
+      "loss": 0.9286,
+      "step": 6936
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.262175789156186e-06,
+      "loss": 0.8867,
+      "step": 6937
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.259431902737195e-06,
+      "loss": 0.8442,
+      "step": 6938
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.256688476613716e-06,
+      "loss": 0.9076,
+      "step": 6939
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2539455110521385e-06,
+      "loss": 0.8696,
+      "step": 6940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.25120300631879e-06,
+      "loss": 0.8567,
+      "step": 6941
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.248460962679958e-06,
+      "loss": 0.8435,
+      "step": 6942
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.245719380401893e-06,
+      "loss": 0.9375,
+      "step": 6943
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.24297825975079e-06,
+      "loss": 0.868,
+      "step": 6944
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.240237600992803e-06,
+      "loss": 0.9537,
+      "step": 6945
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.237497404394044e-06,
+      "loss": 0.8567,
+      "step": 6946
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.234757670220583e-06,
+      "loss": 0.8584,
+      "step": 6947
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.232018398738436e-06,
+      "loss": 0.8551,
+      "step": 6948
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.229279590213574e-06,
+      "loss": 0.866,
+      "step": 6949
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.226541244911936e-06,
+      "loss": 0.8195,
+      "step": 6950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.223803363099402e-06,
+      "loss": 0.8715,
+      "step": 6951
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.221065945041811e-06,
+      "loss": 0.8486,
+      "step": 6952
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.218328991004965e-06,
+      "loss": 0.9083,
+      "step": 6953
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.215592501254609e-06,
+      "loss": 0.8876,
+      "step": 6954
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.212856476056449e-06,
+      "loss": 0.868,
+      "step": 6955
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.210120915676147e-06,
+      "loss": 0.7359,
+      "step": 6956
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.207385820379318e-06,
+      "loss": 0.9022,
+      "step": 6957
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2046511904315265e-06,
+      "loss": 0.9072,
+      "step": 6958
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.201917026098302e-06,
+      "loss": 0.8078,
+      "step": 6959
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.199183327645128e-06,
+      "loss": 0.871,
+      "step": 6960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.196450095337433e-06,
+      "loss": 0.8503,
+      "step": 6961
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.193717329440604e-06,
+      "loss": 0.8909,
+      "step": 6962
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.190985030219991e-06,
+      "loss": 0.8121,
+      "step": 6963
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.188253197940889e-06,
+      "loss": 0.8713,
+      "step": 6964
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1855218328685494e-06,
+      "loss": 0.8409,
+      "step": 6965
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.182790935268185e-06,
+      "loss": 0.9145,
+      "step": 6966
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.180060505404954e-06,
+      "loss": 0.8111,
+      "step": 6967
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.177330543543971e-06,
+      "loss": 0.8841,
+      "step": 6968
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.174601049950315e-06,
+      "loss": 0.817,
+      "step": 6969
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.171872024889004e-06,
+      "loss": 0.9128,
+      "step": 6970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1691434686250204e-06,
+      "loss": 0.8033,
+      "step": 6971
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.166415381423306e-06,
+      "loss": 0.7677,
+      "step": 6972
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.163687763548746e-06,
+      "loss": 0.8673,
+      "step": 6973
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.160960615266179e-06,
+      "loss": 0.8493,
+      "step": 6974
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.15823393684041e-06,
+      "loss": 0.7959,
+      "step": 6975
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.155507728536191e-06,
+      "loss": 0.8324,
+      "step": 6976
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1527819906182215e-06,
+      "loss": 0.9142,
+      "step": 6977
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.150056723351173e-06,
+      "loss": 0.9189,
+      "step": 6978
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.147331926999658e-06,
+      "loss": 0.8954,
+      "step": 6979
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.14460760182824e-06,
+      "loss": 0.8017,
+      "step": 6980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.141883748101448e-06,
+      "loss": 0.8457,
+      "step": 6981
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.139160366083765e-06,
+      "loss": 0.7988,
+      "step": 6982
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.136437456039619e-06,
+      "loss": 0.8265,
+      "step": 6983
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.133715018233393e-06,
+      "loss": 0.8441,
+      "step": 6984
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.130993052929435e-06,
+      "loss": 0.836,
+      "step": 6985
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.128271560392037e-06,
+      "loss": 0.8603,
+      "step": 6986
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1255505408854445e-06,
+      "loss": 0.9017,
+      "step": 6987
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.122829994673866e-06,
+      "loss": 0.8986,
+      "step": 6988
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1201099220214576e-06,
+      "loss": 0.8888,
+      "step": 6989
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.117390323192326e-06,
+      "loss": 0.7976,
+      "step": 6990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.114671198450543e-06,
+      "loss": 0.8434,
+      "step": 6991
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.111952548060126e-06,
+      "loss": 0.8269,
+      "step": 6992
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.109234372285041e-06,
+      "loss": 0.8266,
+      "step": 6993
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.106516671389224e-06,
+      "loss": 0.9454,
+      "step": 6994
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.1037994456365535e-06,
+      "loss": 0.9196,
+      "step": 6995
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.101082695290866e-06,
+      "loss": 0.7858,
+      "step": 6996
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.098366420615942e-06,
+      "loss": 0.8747,
+      "step": 6997
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0956506218755344e-06,
+      "loss": 0.8228,
+      "step": 6998
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.092935299333335e-06,
+      "loss": 0.9746,
+      "step": 6999
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.09022045325299e-06,
+      "loss": 0.8375,
+      "step": 7000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.087506083898111e-06,
+      "loss": 0.9367,
+      "step": 7001
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0847921915322486e-06,
+      "loss": 0.8176,
+      "step": 7002
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.082078776418915e-06,
+      "loss": 0.8802,
+      "step": 7003
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.07936583882158e-06,
+      "loss": 0.8047,
+      "step": 7004
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.076653379003653e-06,
+      "loss": 0.9044,
+      "step": 7005
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.073941397228518e-06,
+      "loss": 0.8445,
+      "step": 7006
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0712298937594884e-06,
+      "loss": 0.8708,
+      "step": 7007
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.068518868859854e-06,
+      "loss": 0.8976,
+      "step": 7008
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0658083227928425e-06,
+      "loss": 0.9287,
+      "step": 7009
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.063098255821637e-06,
+      "loss": 0.8744,
+      "step": 7010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.060388668209384e-06,
+      "loss": 0.9034,
+      "step": 7011
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0576795602191734e-06,
+      "loss": 0.8747,
+      "step": 7012
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.054970932114047e-06,
+      "loss": 0.8361,
+      "step": 7013
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.052262784157014e-06,
+      "loss": 0.9272,
+      "step": 7014
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.049555116611023e-06,
+      "loss": 0.8255,
+      "step": 7015
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.046847929738971e-06,
+      "loss": 0.8367,
+      "step": 7016
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0441412238037375e-06,
+      "loss": 0.8366,
+      "step": 7017
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.041434999068127e-06,
+      "loss": 0.8942,
+      "step": 7018
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0387292557949045e-06,
+      "loss": 0.9054,
+      "step": 7019
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.036023994246787e-06,
+      "loss": 0.8732,
+      "step": 7020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.033319214686455e-06,
+      "loss": 0.8486,
+      "step": 7021
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.030614917376532e-06,
+      "loss": 0.8541,
+      "step": 7022
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.027911102579592e-06,
+      "loss": 0.79,
+      "step": 7023
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.025207770558176e-06,
+      "loss": 0.851,
+      "step": 7024
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.022504921574767e-06,
+      "loss": 0.7903,
+      "step": 7025
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0198025558917985e-06,
+      "loss": 0.8732,
+      "step": 7026
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.017100673771671e-06,
+      "loss": 0.8069,
+      "step": 7027
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.014399275476721e-06,
+      "loss": 0.898,
+      "step": 7028
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.011698361269251e-06,
+      "loss": 0.8678,
+      "step": 7029
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.008997931411517e-06,
+      "loss": 0.851,
+      "step": 7030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.006297986165716e-06,
+      "loss": 0.8186,
+      "step": 7031
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.003598525794002e-06,
+      "loss": 0.8273,
+      "step": 7032
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0008995505584955e-06,
+      "loss": 0.9288,
+      "step": 7033
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.998201060721253e-06,
+      "loss": 0.8458,
+      "step": 7034
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.995503056544286e-06,
+      "loss": 0.8806,
+      "step": 7035
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.992805538289571e-06,
+      "loss": 0.9255,
+      "step": 7036
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.990108506219026e-06,
+      "loss": 0.7553,
+      "step": 7037
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.987411960594521e-06,
+      "loss": 0.7639,
+      "step": 7038
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9847159016778905e-06,
+      "loss": 0.8092,
+      "step": 7039
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.982020329730904e-06,
+      "loss": 0.8986,
+      "step": 7040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.979325245015307e-06,
+      "loss": 0.8498,
+      "step": 7041
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.976630647792771e-06,
+      "loss": 0.8275,
+      "step": 7042
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.973936538324944e-06,
+      "loss": 0.8765,
+      "step": 7043
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.971242916873412e-06,
+      "loss": 0.841,
+      "step": 7044
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.968549783699714e-06,
+      "loss": 0.9015,
+      "step": 7045
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.965857139065354e-06,
+      "loss": 0.8432,
+      "step": 7046
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.963164983231776e-06,
+      "loss": 0.8199,
+      "step": 7047
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9604733164603755e-06,
+      "loss": 0.8057,
+      "step": 7048
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.957782139012515e-06,
+      "loss": 0.8902,
+      "step": 7049
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.955091451149495e-06,
+      "loss": 0.8856,
+      "step": 7050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9524012531325696e-06,
+      "loss": 0.8783,
+      "step": 7051
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9497115452229535e-06,
+      "loss": 0.8729,
+      "step": 7052
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9470223276818145e-06,
+      "loss": 0.8763,
+      "step": 7053
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9443336007702614e-06,
+      "loss": 0.8579,
+      "step": 7054
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.94164536474936e-06,
+      "loss": 0.8637,
+      "step": 7055
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.938957619880138e-06,
+      "loss": 0.8655,
+      "step": 7056
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.936270366423563e-06,
+      "loss": 0.8976,
+      "step": 7057
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9335836046405575e-06,
+      "loss": 0.8622,
+      "step": 7058
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.930897334792004e-06,
+      "loss": 0.8981,
+      "step": 7059
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.928211557138728e-06,
+      "loss": 0.8362,
+      "step": 7060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.925526271941507e-06,
+      "loss": 0.8878,
+      "step": 7061
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.922841479461083e-06,
+      "loss": 0.8413,
+      "step": 7062
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9201571799581346e-06,
+      "loss": 0.8479,
+      "step": 7063
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.917473373693305e-06,
+      "loss": 0.8967,
+      "step": 7064
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.914790060927177e-06,
+      "loss": 0.8637,
+      "step": 7065
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9121072419203016e-06,
+      "loss": 0.8346,
+      "step": 7066
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.909424916933168e-06,
+      "loss": 0.8901,
+      "step": 7067
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.906743086226218e-06,
+      "loss": 0.8807,
+      "step": 7068
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.904061750059858e-06,
+      "loss": 0.8838,
+      "step": 7069
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.901380908694434e-06,
+      "loss": 0.8945,
+      "step": 7070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.898700562390244e-06,
+      "loss": 0.7932,
+      "step": 7071
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8960207114075495e-06,
+      "loss": 0.821,
+      "step": 7072
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8933413560065525e-06,
+      "loss": 0.9012,
+      "step": 7073
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.890662496447407e-06,
+      "loss": 0.9174,
+      "step": 7074
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.887984132990228e-06,
+      "loss": 0.8193,
+      "step": 7075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8853062658950765e-06,
+      "loss": 0.8749,
+      "step": 7076
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.882628895421962e-06,
+      "loss": 0.9036,
+      "step": 7077
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.879952021830856e-06,
+      "loss": 0.7776,
+      "step": 7078
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.877275645381672e-06,
+      "loss": 0.7827,
+      "step": 7079
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.874599766334276e-06,
+      "loss": 0.8961,
+      "step": 7080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8719243849484866e-06,
+      "loss": 0.8735,
+      "step": 7081
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8692495014840825e-06,
+      "loss": 0.9637,
+      "step": 7082
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.866575116200783e-06,
+      "loss": 0.8036,
+      "step": 7083
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.863901229358261e-06,
+      "loss": 0.8868,
+      "step": 7084
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.86122784121615e-06,
+      "loss": 0.8553,
+      "step": 7085
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.858554952034019e-06,
+      "loss": 0.861,
+      "step": 7086
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.855882562071404e-06,
+      "loss": 0.872,
+      "step": 7087
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.853210671587789e-06,
+      "loss": 0.8802,
+      "step": 7088
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8505392808426036e-06,
+      "loss": 0.9183,
+      "step": 7089
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.847868390095227e-06,
+      "loss": 0.8182,
+      "step": 7090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.845197999605005e-06,
+      "loss": 0.8731,
+      "step": 7091
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.842528109631218e-06,
+      "loss": 0.8293,
+      "step": 7092
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.839858720433103e-06,
+      "loss": 0.9091,
+      "step": 7093
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.837189832269858e-06,
+      "loss": 0.8201,
+      "step": 7094
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.834521445400621e-06,
+      "loss": 0.8468,
+      "step": 7095
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8318535600844775e-06,
+      "loss": 0.7918,
+      "step": 7096
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.829186176580481e-06,
+      "loss": 0.8986,
+      "step": 7097
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8265192951476206e-06,
+      "loss": 0.8826,
+      "step": 7098
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.823852916044849e-06,
+      "loss": 0.8437,
+      "step": 7099
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8211870395310556e-06,
+      "loss": 0.8065,
+      "step": 7100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.818521665865098e-06,
+      "loss": 0.8767,
+      "step": 7101
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.815856795305772e-06,
+      "loss": 0.898,
+      "step": 7102
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.813192428111826e-06,
+      "loss": 0.8487,
+      "step": 7103
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.81052856454197e-06,
+      "loss": 0.9145,
+      "step": 7104
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.807865204854852e-06,
+      "loss": 0.8835,
+      "step": 7105
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.805202349309074e-06,
+      "loss": 0.8999,
+      "step": 7106
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.802539998163199e-06,
+      "loss": 0.8896,
+      "step": 7107
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7998781516757295e-06,
+      "loss": 0.8322,
+      "step": 7108
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.797216810105121e-06,
+      "loss": 0.8945,
+      "step": 7109
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.794555973709783e-06,
+      "loss": 0.8901,
+      "step": 7110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.791895642748081e-06,
+      "loss": 0.88,
+      "step": 7111
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.789235817478322e-06,
+      "loss": 0.8335,
+      "step": 7112
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.786576498158761e-06,
+      "loss": 0.8279,
+      "step": 7113
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.783917685047621e-06,
+      "loss": 0.8759,
+      "step": 7114
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.781259378403059e-06,
+      "loss": 0.8536,
+      "step": 7115
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.778601578483187e-06,
+      "loss": 0.872,
+      "step": 7116
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775944285546076e-06,
+      "loss": 0.8814,
+      "step": 7117
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.773287499849737e-06,
+      "loss": 0.8382,
+      "step": 7118
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.770631221652134e-06,
+      "loss": 0.8597,
+      "step": 7119
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.767975451211191e-06,
+      "loss": 0.9103,
+      "step": 7120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.765320188784769e-06,
+      "loss": 0.9479,
+      "step": 7121
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.762665434630692e-06,
+      "loss": 0.845,
+      "step": 7122
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.760011189006722e-06,
+      "loss": 0.8478,
+      "step": 7123
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.757357452170588e-06,
+      "loss": 0.84,
+      "step": 7124
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.754704224379953e-06,
+      "loss": 0.8171,
+      "step": 7125
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.752051505892438e-06,
+      "loss": 0.927,
+      "step": 7126
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.749399296965619e-06,
+      "loss": 0.8807,
+      "step": 7127
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.746747597857014e-06,
+      "loss": 0.882,
+      "step": 7128
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.744096408824095e-06,
+      "loss": 0.8753,
+      "step": 7129
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.741445730124287e-06,
+      "loss": 0.9025,
+      "step": 7130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7387955620149654e-06,
+      "loss": 0.861,
+      "step": 7131
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.736145904753445e-06,
+      "loss": 0.8696,
+      "step": 7132
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.733496758597007e-06,
+      "loss": 0.9301,
+      "step": 7133
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.730848123802877e-06,
+      "loss": 0.8054,
+      "step": 7134
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.728200000628225e-06,
+      "loss": 0.9122,
+      "step": 7135
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.725552389330183e-06,
+      "loss": 0.8454,
+      "step": 7136
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.72290529016582e-06,
+      "loss": 0.8888,
+      "step": 7137
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.720258703392161e-06,
+      "loss": 0.8195,
+      "step": 7138
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.71761262926619e-06,
+      "loss": 0.912,
+      "step": 7139
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.714967068044826e-06,
+      "loss": 0.9178,
+      "step": 7140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.712322019984947e-06,
+      "loss": 0.8479,
+      "step": 7141
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7096774853433765e-06,
+      "loss": 0.845,
+      "step": 7142
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7070334643769e-06,
+      "loss": 0.8737,
+      "step": 7143
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.704389957342237e-06,
+      "loss": 0.9464,
+      "step": 7144
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.70174696449606e-06,
+      "loss": 0.7551,
+      "step": 7145
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.699104486095008e-06,
+      "loss": 0.8055,
+      "step": 7146
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.696462522395654e-06,
+      "loss": 0.8828,
+      "step": 7147
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.69382107365452e-06,
+      "loss": 0.8997,
+      "step": 7148
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.691180140128089e-06,
+      "loss": 0.8428,
+      "step": 7149
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6885397220727855e-06,
+      "loss": 0.8904,
+      "step": 7150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.685899819744983e-06,
+      "loss": 0.8729,
+      "step": 7151
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.683260433401016e-06,
+      "loss": 0.8678,
+      "step": 7152
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.680621563297157e-06,
+      "loss": 0.8758,
+      "step": 7153
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.677983209689631e-06,
+      "loss": 0.9044,
+      "step": 7154
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.675345372834618e-06,
+      "loss": 0.8866,
+      "step": 7155
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6727080529882394e-06,
+      "loss": 0.8235,
+      "step": 7156
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.670071250406579e-06,
+      "loss": 0.871,
+      "step": 7157
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.667434965345654e-06,
+      "loss": 0.848,
+      "step": 7158
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.664799198061449e-06,
+      "loss": 0.842,
+      "step": 7159
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6621639488098856e-06,
+      "loss": 0.8913,
+      "step": 7160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.659529217846834e-06,
+      "loss": 0.7794,
+      "step": 7161
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.656895005428127e-06,
+      "loss": 0.8966,
+      "step": 7162
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.654261311809536e-06,
+      "loss": 0.8792,
+      "step": 7163
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.651628137246781e-06,
+      "loss": 0.8624,
+      "step": 7164
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6489954819955414e-06,
+      "loss": 0.8421,
+      "step": 7165
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6463633463114395e-06,
+      "loss": 0.8698,
+      "step": 7166
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.643731730450043e-06,
+      "loss": 0.9021,
+      "step": 7167
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.641100634666877e-06,
+      "loss": 0.9293,
+      "step": 7168
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6384700592174194e-06,
+      "loss": 0.9108,
+      "step": 7169
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.635840004357086e-06,
+      "loss": 0.9118,
+      "step": 7170
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.633210470341244e-06,
+      "loss": 0.851,
+      "step": 7171
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.630581457425222e-06,
+      "loss": 0.8478,
+      "step": 7172
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.627952965864284e-06,
+      "loss": 0.8815,
+      "step": 7173
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.625324995913648e-06,
+      "loss": 0.8444,
+      "step": 7174
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6226975478284875e-06,
+      "loss": 0.8157,
+      "step": 7175
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.620070621863917e-06,
+      "loss": 0.8514,
+      "step": 7176
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.617444218274999e-06,
+      "loss": 0.8285,
+      "step": 7177
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.614818337316759e-06,
+      "loss": 0.8106,
+      "step": 7178
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.612192979244158e-06,
+      "loss": 0.8858,
+      "step": 7179
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.609568144312107e-06,
+      "loss": 0.8305,
+      "step": 7180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.606943832775473e-06,
+      "loss": 0.9185,
+      "step": 7181
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6043200448890724e-06,
+      "loss": 0.9304,
+      "step": 7182
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.601696780907665e-06,
+      "loss": 0.8817,
+      "step": 7183
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.599074041085958e-06,
+      "loss": 0.8624,
+      "step": 7184
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.596451825678619e-06,
+      "loss": 0.864,
+      "step": 7185
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.593830134940256e-06,
+      "loss": 0.8719,
+      "step": 7186
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.59120896912542e-06,
+      "loss": 0.8275,
+      "step": 7187
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.588588328488629e-06,
+      "loss": 0.8766,
+      "step": 7188
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.585968213284335e-06,
+      "loss": 0.8389,
+      "step": 7189
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5833486237669414e-06,
+      "loss": 0.9147,
+      "step": 7190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.580729560190803e-06,
+      "loss": 0.8295,
+      "step": 7191
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.578111022810231e-06,
+      "loss": 0.9031,
+      "step": 7192
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.575493011879467e-06,
+      "loss": 0.849,
+      "step": 7193
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5728755276527225e-06,
+      "loss": 0.8915,
+      "step": 7194
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5702585703841425e-06,
+      "loss": 0.8247,
+      "step": 7195
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.567642140327823e-06,
+      "loss": 0.8397,
+      "step": 7196
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.565026237737819e-06,
+      "loss": 0.9641,
+      "step": 7197
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.562410862868123e-06,
+      "loss": 0.8218,
+      "step": 7198
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.559796015972677e-06,
+      "loss": 0.9045,
+      "step": 7199
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.557181697305383e-06,
+      "loss": 0.8781,
+      "step": 7200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.554567907120079e-06,
+      "loss": 0.8317,
+      "step": 7201
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.551954645670557e-06,
+      "loss": 0.8865,
+      "step": 7202
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.549341913210551e-06,
+      "loss": 0.8194,
+      "step": 7203
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.546729709993762e-06,
+      "loss": 0.8411,
+      "step": 7204
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.544118036273822e-06,
+      "loss": 0.89,
+      "step": 7205
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.541506892304314e-06,
+      "loss": 0.9042,
+      "step": 7206
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.538896278338777e-06,
+      "loss": 0.9387,
+      "step": 7207
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.536286194630694e-06,
+      "loss": 0.8446,
+      "step": 7208
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.53367664143349e-06,
+      "loss": 0.8945,
+      "step": 7209
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.531067619000553e-06,
+      "loss": 0.8917,
+      "step": 7210
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5284591275852096e-06,
+      "loss": 0.913,
+      "step": 7211
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.525851167440731e-06,
+      "loss": 0.8123,
+      "step": 7212
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5232437388203525e-06,
+      "loss": 0.8942,
+      "step": 7213
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.52063684197724e-06,
+      "loss": 0.8431,
+      "step": 7214
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.518030477164518e-06,
+      "loss": 0.8671,
+      "step": 7215
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.515424644635254e-06,
+      "loss": 0.8403,
+      "step": 7216
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.512819344642477e-06,
+      "loss": 0.7833,
+      "step": 7217
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.510214577439146e-06,
+      "loss": 0.8073,
+      "step": 7218
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.507610343278174e-06,
+      "loss": 0.8684,
+      "step": 7219
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5050066424124324e-06,
+      "loss": 0.8144,
+      "step": 7220
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.502403475094729e-06,
+      "loss": 0.8032,
+      "step": 7221
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.49980084157782e-06,
+      "loss": 0.8978,
+      "step": 7222
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.497198742114421e-06,
+      "loss": 0.8681,
+      "step": 7223
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.494597176957186e-06,
+      "loss": 0.8283,
+      "step": 7224
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.491996146358714e-06,
+      "loss": 0.8451,
+      "step": 7225
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.489395650571562e-06,
+      "loss": 0.8406,
+      "step": 7226
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.486795689848235e-06,
+      "loss": 0.8892,
+      "step": 7227
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4841962644411765e-06,
+      "loss": 0.8567,
+      "step": 7228
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.48159737460278e-06,
+      "loss": 0.8096,
+      "step": 7229
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4789990205854e-06,
+      "loss": 0.8236,
+      "step": 7230
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.476401202641322e-06,
+      "loss": 0.827,
+      "step": 7231
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.473803921022784e-06,
+      "loss": 0.8184,
+      "step": 7232
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.471207175981983e-06,
+      "loss": 0.8073,
+      "step": 7233
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.468610967771051e-06,
+      "loss": 0.8616,
+      "step": 7234
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.466015296642069e-06,
+      "loss": 0.8872,
+      "step": 7235
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4634201628470766e-06,
+      "loss": 0.8811,
+      "step": 7236
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.46082556663805e-06,
+      "loss": 0.9312,
+      "step": 7237
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.458231508266912e-06,
+      "loss": 0.7808,
+      "step": 7238
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.455637987985545e-06,
+      "loss": 0.8442,
+      "step": 7239
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.453045006045773e-06,
+      "loss": 0.8964,
+      "step": 7240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.450452562699366e-06,
+      "loss": 0.9222,
+      "step": 7241
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.447860658198035e-06,
+      "loss": 0.8883,
+      "step": 7242
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.445269292793459e-06,
+      "loss": 0.8393,
+      "step": 7243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.442678466737245e-06,
+      "loss": 0.8502,
+      "step": 7244
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.440088180280952e-06,
+      "loss": 0.8625,
+      "step": 7245
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4374984336760975e-06,
+      "loss": 0.9024,
+      "step": 7246
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.434909227174133e-06,
+      "loss": 0.853,
+      "step": 7247
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.432320561026461e-06,
+      "loss": 0.8748,
+      "step": 7248
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.42973243548444e-06,
+      "loss": 0.7752,
+      "step": 7249
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.427144850799363e-06,
+      "loss": 0.8636,
+      "step": 7250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.424557807222479e-06,
+      "loss": 0.8717,
+      "step": 7251
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.421971305004989e-06,
+      "loss": 0.823,
+      "step": 7252
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.419385344398028e-06,
+      "loss": 0.829,
+      "step": 7253
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.416799925652684e-06,
+      "loss": 0.9565,
+      "step": 7254
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.41421504902e-06,
+      "loss": 0.8784,
+      "step": 7255
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.411630714750956e-06,
+      "loss": 0.8759,
+      "step": 7256
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.40904692309648e-06,
+      "loss": 0.7508,
+      "step": 7257
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4064636743074605e-06,
+      "loss": 0.9336,
+      "step": 7258
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.403880968634716e-06,
+      "loss": 0.8848,
+      "step": 7259
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.40129880632902e-06,
+      "loss": 0.8934,
+      "step": 7260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.398717187641093e-06,
+      "loss": 0.915,
+      "step": 7261
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.396136112821608e-06,
+      "loss": 0.8497,
+      "step": 7262
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.393555582121177e-06,
+      "loss": 0.8891,
+      "step": 7263
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.390975595790358e-06,
+      "loss": 0.8554,
+      "step": 7264
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3883961540796664e-06,
+      "loss": 0.858,
+      "step": 7265
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.385817257239556e-06,
+      "loss": 0.8362,
+      "step": 7266
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.383238905520427e-06,
+      "loss": 0.9008,
+      "step": 7267
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.380661099172636e-06,
+      "loss": 0.8586,
+      "step": 7268
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.378083838446478e-06,
+      "loss": 0.8331,
+      "step": 7269
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.375507123592194e-06,
+      "loss": 0.8996,
+      "step": 7270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.372930954859982e-06,
+      "loss": 0.8253,
+      "step": 7271
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.370355332499977e-06,
+      "loss": 0.8735,
+      "step": 7272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.367780256762262e-06,
+      "loss": 0.8079,
+      "step": 7273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.365205727896872e-06,
+      "loss": 0.8791,
+      "step": 7274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.36263174615379e-06,
+      "loss": 0.9646,
+      "step": 7275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.36005831178294e-06,
+      "loss": 0.8849,
+      "step": 7276
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.357485425034189e-06,
+      "loss": 0.9139,
+      "step": 7277
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.354913086157367e-06,
+      "loss": 0.7977,
+      "step": 7278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.352341295402233e-06,
+      "loss": 0.8028,
+      "step": 7279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.349770053018502e-06,
+      "loss": 0.8428,
+      "step": 7280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.347199359255837e-06,
+      "loss": 0.8958,
+      "step": 7281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.344629214363845e-06,
+      "loss": 0.8206,
+      "step": 7282
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.342059618592073e-06,
+      "loss": 0.8272,
+      "step": 7283
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.339490572190031e-06,
+      "loss": 0.8628,
+      "step": 7284
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3369220754071565e-06,
+      "loss": 0.8371,
+      "step": 7285
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.334354128492851e-06,
+      "loss": 0.8485,
+      "step": 7286
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.331786731696449e-06,
+      "loss": 0.8736,
+      "step": 7287
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.329219885267244e-06,
+      "loss": 0.8987,
+      "step": 7288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.326653589454465e-06,
+      "loss": 0.859,
+      "step": 7289
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.324087844507289e-06,
+      "loss": 0.8349,
+      "step": 7290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.321522650674849e-06,
+      "loss": 0.812,
+      "step": 7291
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.318958008206214e-06,
+      "loss": 0.8463,
+      "step": 7292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.316393917350402e-06,
+      "loss": 0.8371,
+      "step": 7293
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.313830378356384e-06,
+      "loss": 0.8656,
+      "step": 7294
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.311267391473069e-06,
+      "loss": 0.8831,
+      "step": 7295
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3087049569493136e-06,
+      "loss": 0.9421,
+      "step": 7296
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.306143075033924e-06,
+      "loss": 0.8248,
+      "step": 7297
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303581745975656e-06,
+      "loss": 0.8841,
+      "step": 7298
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.301020970023204e-06,
+      "loss": 0.911,
+      "step": 7299
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2984607474252084e-06,
+      "loss": 0.8927,
+      "step": 7300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.295901078430267e-06,
+      "loss": 0.8997,
+      "step": 7301
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.293341963286912e-06,
+      "loss": 0.8642,
+      "step": 7302
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.290783402243622e-06,
+      "loss": 0.8892,
+      "step": 7303
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.288225395548835e-06,
+      "loss": 0.8878,
+      "step": 7304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.285667943450921e-06,
+      "loss": 0.8854,
+      "step": 7305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.283111046198198e-06,
+      "loss": 0.8719,
+      "step": 7306
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.280554704038941e-06,
+      "loss": 0.817,
+      "step": 7307
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.277998917221354e-06,
+      "loss": 0.8674,
+      "step": 7308
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.275443685993604e-06,
+      "loss": 0.7742,
+      "step": 7309
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.272889010603798e-06,
+      "loss": 0.9053,
+      "step": 7310
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2703348912999835e-06,
+      "loss": 0.8036,
+      "step": 7311
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.267781328330155e-06,
+      "loss": 0.8628,
+      "step": 7312
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.265228321942264e-06,
+      "loss": 0.8526,
+      "step": 7313
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.262675872384197e-06,
+      "loss": 0.8355,
+      "step": 7314
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.260123979903783e-06,
+      "loss": 0.8759,
+      "step": 7315
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.257572644748813e-06,
+      "loss": 0.8545,
+      "step": 7316
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.255021867167011e-06,
+      "loss": 0.8282,
+      "step": 7317
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.252471647406045e-06,
+      "loss": 0.8408,
+      "step": 7318
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.249921985713543e-06,
+      "loss": 0.8045,
+      "step": 7319
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2473728823370605e-06,
+      "loss": 0.8718,
+      "step": 7320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.244824337524117e-06,
+      "loss": 0.809,
+      "step": 7321
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.242276351522161e-06,
+      "loss": 0.9148,
+      "step": 7322
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.239728924578602e-06,
+      "loss": 0.9241,
+      "step": 7323
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.237182056940784e-06,
+      "loss": 0.8325,
+      "step": 7324
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.234635748855998e-06,
+      "loss": 0.8827,
+      "step": 7325
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.232090000571488e-06,
+      "loss": 0.9575,
+      "step": 7326
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.229544812334438e-06,
+      "loss": 0.8698,
+      "step": 7327
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2270001843919714e-06,
+      "loss": 0.8174,
+      "step": 7328
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.224456116991175e-06,
+      "loss": 0.868,
+      "step": 7329
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.221912610379065e-06,
+      "loss": 0.8207,
+      "step": 7330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.219369664802605e-06,
+      "loss": 0.7612,
+      "step": 7331
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216827280508712e-06,
+      "loss": 0.8585,
+      "step": 7332
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214285457744247e-06,
+      "loss": 0.8768,
+      "step": 7333
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211744196756011e-06,
+      "loss": 0.8239,
+      "step": 7334
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209203497790749e-06,
+      "loss": 0.8079,
+      "step": 7335
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.206663361095164e-06,
+      "loss": 0.9075,
+      "step": 7336
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.204123786915889e-06,
+      "loss": 0.8391,
+      "step": 7337
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.201584775499509e-06,
+      "loss": 0.774,
+      "step": 7338
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.199046327092561e-06,
+      "loss": 0.7978,
+      "step": 7339
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.196508441941516e-06,
+      "loss": 0.7321,
+      "step": 7340
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.193971120292793e-06,
+      "loss": 0.8943,
+      "step": 7341
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.191434362392768e-06,
+      "loss": 0.91,
+      "step": 7342
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.188898168487742e-06,
+      "loss": 0.85,
+      "step": 7343
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.186362538823981e-06,
+      "loss": 0.8068,
+      "step": 7344
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.183827473647679e-06,
+      "loss": 0.8255,
+      "step": 7345
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.181292973204992e-06,
+      "loss": 0.851,
+      "step": 7346
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.178759037742009e-06,
+      "loss": 0.8496,
+      "step": 7347
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1762256675047655e-06,
+      "loss": 0.8945,
+      "step": 7348
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.173692862739248e-06,
+      "loss": 0.8693,
+      "step": 7349
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.171160623691384e-06,
+      "loss": 0.8503,
+      "step": 7350
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.168628950607042e-06,
+      "loss": 0.8748,
+      "step": 7351
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.166097843732048e-06,
+      "loss": 0.8735,
+      "step": 7352
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.16356730331216e-06,
+      "loss": 0.8568,
+      "step": 7353
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.161037329593085e-06,
+      "loss": 0.8208,
+      "step": 7354
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.158507922820478e-06,
+      "loss": 0.8708,
+      "step": 7355
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.155979083239942e-06,
+      "loss": 0.8885,
+      "step": 7356
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.153450811097014e-06,
+      "loss": 0.7847,
+      "step": 7357
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1509231066371815e-06,
+      "loss": 0.8521,
+      "step": 7358
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.148395970105882e-06,
+      "loss": 0.8752,
+      "step": 7359
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1458694017484915e-06,
+      "loss": 0.8815,
+      "step": 7360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.143343401810327e-06,
+      "loss": 0.8798,
+      "step": 7361
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.140817970536664e-06,
+      "loss": 0.8726,
+      "step": 7362
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1382931081727115e-06,
+      "loss": 0.8661,
+      "step": 7363
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.135768814963622e-06,
+      "loss": 0.8441,
+      "step": 7364
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1332450911545054e-06,
+      "loss": 0.8519,
+      "step": 7365
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.130721936990399e-06,
+      "loss": 0.8013,
+      "step": 7366
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.128199352716299e-06,
+      "loss": 0.8352,
+      "step": 7367
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1256773385771444e-06,
+      "loss": 0.9182,
+      "step": 7368
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.12315589481781e-06,
+      "loss": 0.8144,
+      "step": 7369
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.120635021683122e-06,
+      "loss": 0.8783,
+      "step": 7370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.118114719417851e-06,
+      "loss": 0.892,
+      "step": 7371
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.115594988266711e-06,
+      "loss": 0.8845,
+      "step": 7372
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.113075828474357e-06,
+      "loss": 0.9288,
+      "step": 7373
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1105572402853976e-06,
+      "loss": 0.8418,
+      "step": 7374
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.108039223944379e-06,
+      "loss": 0.8757,
+      "step": 7375
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1055217796957895e-06,
+      "loss": 0.7518,
+      "step": 7376
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.103004907784071e-06,
+      "loss": 0.8574,
+      "step": 7377
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.100488608453599e-06,
+      "loss": 0.9132,
+      "step": 7378
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0979728819487055e-06,
+      "loss": 0.8287,
+      "step": 7379
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.095457728513652e-06,
+      "loss": 0.9872,
+      "step": 7380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.092943148392663e-06,
+      "loss": 0.8723,
+      "step": 7381
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.09042914182989e-06,
+      "loss": 0.7894,
+      "step": 7382
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.087915709069434e-06,
+      "loss": 0.8346,
+      "step": 7383
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.08540285035535e-06,
+      "loss": 0.9371,
+      "step": 7384
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.082890565931624e-06,
+      "loss": 0.8281,
+      "step": 7385
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0803788560421885e-06,
+      "loss": 0.9182,
+      "step": 7386
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.077867720930932e-06,
+      "loss": 0.8232,
+      "step": 7387
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.075357160841671e-06,
+      "loss": 0.9463,
+      "step": 7388
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.072847176018174e-06,
+      "loss": 0.8689,
+      "step": 7389
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.070337766704155e-06,
+      "loss": 0.8392,
+      "step": 7390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0678289331432765e-06,
+      "loss": 0.826,
+      "step": 7391
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.065320675579132e-06,
+      "loss": 0.8929,
+      "step": 7392
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.062812994255264e-06,
+      "loss": 0.9258,
+      "step": 7393
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0603058894151685e-06,
+      "loss": 0.8754,
+      "step": 7394
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.057799361302274e-06,
+      "loss": 0.8342,
+      "step": 7395
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.055293410159954e-06,
+      "loss": 0.9175,
+      "step": 7396
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.052788036231536e-06,
+      "loss": 0.8871,
+      "step": 7397
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.050283239760282e-06,
+      "loss": 0.8605,
+      "step": 7398
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0477790209893965e-06,
+      "loss": 0.8816,
+      "step": 7399
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.045275380162038e-06,
+      "loss": 0.9,
+      "step": 7400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.042772317521296e-06,
+      "loss": 0.8982,
+      "step": 7401
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.04026983331022e-06,
+      "loss": 0.8734,
+      "step": 7402
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.037767927771784e-06,
+      "loss": 0.8716,
+      "step": 7403
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.035266601148924e-06,
+      "loss": 0.9117,
+      "step": 7404
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.03276585368451e-06,
+      "loss": 0.8791,
+      "step": 7405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.03026568562135e-06,
+      "loss": 0.8635,
+      "step": 7406
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.027766097202213e-06,
+      "loss": 0.8228,
+      "step": 7407
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.025267088669797e-06,
+      "loss": 0.8723,
+      "step": 7408
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.022768660266745e-06,
+      "loss": 0.8129,
+      "step": 7409
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.020270812235656e-06,
+      "loss": 0.824,
+      "step": 7410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.017773544819059e-06,
+      "loss": 0.8381,
+      "step": 7411
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.015276858259427e-06,
+      "loss": 0.8182,
+      "step": 7412
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.012780752799187e-06,
+      "loss": 0.9013,
+      "step": 7413
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.010285228680705e-06,
+      "loss": 0.8868,
+      "step": 7414
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.007790286146284e-06,
+      "loss": 0.9054,
+      "step": 7415
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.005295925438181e-06,
+      "loss": 0.9346,
+      "step": 7416
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0028021467985886e-06,
+      "loss": 0.8067,
+      "step": 7417
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.000308950469646e-06,
+      "loss": 0.8629,
+      "step": 7418
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.997816336693431e-06,
+      "loss": 0.8189,
+      "step": 7419
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.995324305711976e-06,
+      "loss": 0.8966,
+      "step": 7420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.992832857767247e-06,
+      "loss": 0.8134,
+      "step": 7421
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.990341993101154e-06,
+      "loss": 0.8606,
+      "step": 7422
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9878517119555584e-06,
+      "loss": 0.8577,
+      "step": 7423
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.985362014572256e-06,
+      "loss": 0.8581,
+      "step": 7424
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.982872901192983e-06,
+      "loss": 0.847,
+      "step": 7425
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9803843720594385e-06,
+      "loss": 0.839,
+      "step": 7426
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.977896427413245e-06,
+      "loss": 0.8117,
+      "step": 7427
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.97540906749597e-06,
+      "loss": 0.8335,
+      "step": 7428
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.972922292549137e-06,
+      "loss": 0.8287,
+      "step": 7429
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.970436102814203e-06,
+      "loss": 0.7991,
+      "step": 7430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.967950498532562e-06,
+      "loss": 0.9373,
+      "step": 7431
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.965465479945569e-06,
+      "loss": 0.8544,
+      "step": 7432
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.962981047294509e-06,
+      "loss": 0.8834,
+      "step": 7433
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9604972008206085e-06,
+      "loss": 0.8609,
+      "step": 7434
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.958013940765049e-06,
+      "loss": 0.7977,
+      "step": 7435
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.955531267368942e-06,
+      "loss": 0.8627,
+      "step": 7436
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.953049180873353e-06,
+      "loss": 0.8911,
+      "step": 7437
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.950567681519279e-06,
+      "loss": 0.8273,
+      "step": 7438
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9480867695476755e-06,
+      "loss": 0.9185,
+      "step": 7439
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.945606445199427e-06,
+      "loss": 0.8467,
+      "step": 7440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.943126708715361e-06,
+      "loss": 0.8652,
+      "step": 7441
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.940647560336262e-06,
+      "loss": 0.8553,
+      "step": 7442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.938169000302843e-06,
+      "loss": 0.7843,
+      "step": 7443
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.935691028855763e-06,
+      "loss": 0.8996,
+      "step": 7444
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9332136462356315e-06,
+      "loss": 0.8382,
+      "step": 7445
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.930736852682993e-06,
+      "loss": 0.8144,
+      "step": 7446
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9282606484383325e-06,
+      "loss": 0.8683,
+      "step": 7447
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9257850337420856e-06,
+      "loss": 0.8697,
+      "step": 7448
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9233100088346325e-06,
+      "loss": 0.827,
+      "step": 7449
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.920835573956285e-06,
+      "loss": 0.8441,
+      "step": 7450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.918361729347303e-06,
+      "loss": 0.9176,
+      "step": 7451
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.915888475247894e-06,
+      "loss": 0.825,
+      "step": 7452
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9134158118982034e-06,
+      "loss": 0.8363,
+      "step": 7453
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.910943739538313e-06,
+      "loss": 0.8513,
+      "step": 7454
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.908472258408262e-06,
+      "loss": 0.8567,
+      "step": 7455
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.906001368748023e-06,
+      "loss": 0.9666,
+      "step": 7456
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.903531070797504e-06,
+      "loss": 0.9281,
+      "step": 7457
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.901061364796574e-06,
+      "loss": 0.8353,
+      "step": 7458
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8985922509850314e-06,
+      "loss": 0.8258,
+      "step": 7459
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8961237296026155e-06,
+      "loss": 0.8321,
+      "step": 7460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8936558008890156e-06,
+      "loss": 0.9122,
+      "step": 7461
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.891188465083865e-06,
+      "loss": 0.8719,
+      "step": 7462
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8887217224267305e-06,
+      "loss": 0.8342,
+      "step": 7463
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.886255573157121e-06,
+      "loss": 0.8788,
+      "step": 7464
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8837900175145036e-06,
+      "loss": 0.8844,
+      "step": 7465
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.88132505573827e-06,
+      "loss": 0.9121,
+      "step": 7466
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.878860688067758e-06,
+      "loss": 0.8369,
+      "step": 7467
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.876396914742258e-06,
+      "loss": 0.8099,
+      "step": 7468
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.87393373600099e-06,
+      "loss": 0.8508,
+      "step": 7469
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.871471152083121e-06,
+      "loss": 0.8484,
+      "step": 7470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.869009163227761e-06,
+      "loss": 0.7522,
+      "step": 7471
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.866547769673968e-06,
+      "loss": 0.9277,
+      "step": 7472
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.86408697166073e-06,
+      "loss": 0.8233,
+      "step": 7473
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.861626769426988e-06,
+      "loss": 0.9398,
+      "step": 7474
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.859167163211618e-06,
+      "loss": 0.907,
+      "step": 7475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8567081532534374e-06,
+      "loss": 0.8497,
+      "step": 7476
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.854249739791216e-06,
+      "loss": 0.7827,
+      "step": 7477
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.851791923063655e-06,
+      "loss": 0.9268,
+      "step": 7478
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.849334703309401e-06,
+      "loss": 0.9616,
+      "step": 7479
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.846878080767039e-06,
+      "loss": 0.8846,
+      "step": 7480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.844422055675109e-06,
+      "loss": 0.8466,
+      "step": 7481
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.841966628272079e-06,
+      "loss": 0.8742,
+      "step": 7482
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.839511798796357e-06,
+      "loss": 0.8799,
+      "step": 7483
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.837057567486314e-06,
+      "loss": 0.8777,
+      "step": 7484
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.834603934580241e-06,
+      "loss": 0.869,
+      "step": 7485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.832150900316377e-06,
+      "loss": 0.8976,
+      "step": 7486
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8296984649329105e-06,
+      "loss": 0.8559,
+      "step": 7487
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.827246628667962e-06,
+      "loss": 0.8033,
+      "step": 7488
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8247953917595955e-06,
+      "loss": 0.9137,
+      "step": 7489
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.822344754445826e-06,
+      "loss": 0.9228,
+      "step": 7490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.819894716964597e-06,
+      "loss": 0.8403,
+      "step": 7491
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.817445279553801e-06,
+      "loss": 0.8251,
+      "step": 7492
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.814996442451274e-06,
+      "loss": 0.846,
+      "step": 7493
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8125482058947905e-06,
+      "loss": 0.877,
+      "step": 7494
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.810100570122063e-06,
+      "loss": 0.8859,
+      "step": 7495
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8076535353707523e-06,
+      "loss": 0.8305,
+      "step": 7496
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.805207101878464e-06,
+      "loss": 0.7835,
+      "step": 7497
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8027612698827344e-06,
+      "loss": 0.8051,
+      "step": 7498
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.800316039621044e-06,
+      "loss": 0.8538,
+      "step": 7499
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7978714113308246e-06,
+      "loss": 0.8337,
+      "step": 7500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7954273852494383e-06,
+      "loss": 0.8793,
+      "step": 7501
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7929839616141917e-06,
+      "loss": 0.8656,
+      "step": 7502
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7905411406623384e-06,
+      "loss": 0.8017,
+      "step": 7503
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.788098922631067e-06,
+      "loss": 0.8428,
+      "step": 7504
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7856573077575078e-06,
+      "loss": 0.8884,
+      "step": 7505
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7832162962787355e-06,
+      "loss": 0.7763,
+      "step": 7506
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.78077588843177e-06,
+      "loss": 0.9084,
+      "step": 7507
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7783360844535653e-06,
+      "loss": 0.9084,
+      "step": 7508
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.775896884581014e-06,
+      "loss": 0.9027,
+      "step": 7509
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.773458289050963e-06,
+      "loss": 0.8479,
+      "step": 7510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.771020298100191e-06,
+      "loss": 0.9209,
+      "step": 7511
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.768582911965414e-06,
+      "loss": 0.8335,
+      "step": 7512
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7661461308833034e-06,
+      "loss": 0.8941,
+      "step": 7513
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.763709955090461e-06,
+      "loss": 0.8547,
+      "step": 7514
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7612743848234267e-06,
+      "loss": 0.8403,
+      "step": 7515
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7588394203186963e-06,
+      "loss": 0.8806,
+      "step": 7516
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7564050618126936e-06,
+      "loss": 0.8713,
+      "step": 7517
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.753971309541784e-06,
+      "loss": 0.8394,
+      "step": 7518
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.751538163742282e-06,
+      "loss": 0.8646,
+      "step": 7519
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7491056246504433e-06,
+      "loss": 0.8517,
+      "step": 7520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7466736925024548e-06,
+      "loss": 0.8361,
+      "step": 7521
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7442423675344474e-06,
+      "loss": 0.862,
+      "step": 7522
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.741811649982504e-06,
+      "loss": 0.8592,
+      "step": 7523
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.739381540082635e-06,
+      "loss": 0.823,
+      "step": 7524
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.736952038070795e-06,
+      "loss": 0.7905,
+      "step": 7525
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7345231441828876e-06,
+      "loss": 0.7896,
+      "step": 7526
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.732094858654749e-06,
+      "loss": 0.8517,
+      "step": 7527
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.729667181722154e-06,
+      "loss": 0.853,
+      "step": 7528
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7272401136208313e-06,
+      "loss": 0.8841,
+      "step": 7529
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7248136545864345e-06,
+      "loss": 0.8625,
+      "step": 7530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7223878048545694e-06,
+      "loss": 0.87,
+      "step": 7531
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.719962564660783e-06,
+      "loss": 0.8928,
+      "step": 7532
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7175379342405537e-06,
+      "loss": 0.8135,
+      "step": 7533
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7151139138293056e-06,
+      "loss": 0.8868,
+      "step": 7534
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.712690503662408e-06,
+      "loss": 0.8724,
+      "step": 7535
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7102677039751667e-06,
+      "loss": 0.8354,
+      "step": 7536
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7078455150028227e-06,
+      "loss": 0.9151,
+      "step": 7537
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.705423936980572e-06,
+      "loss": 0.8425,
+      "step": 7538
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7030029701435377e-06,
+      "loss": 0.889,
+      "step": 7539
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.700582614726791e-06,
+      "loss": 0.9099,
+      "step": 7540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6981628709653327e-06,
+      "loss": 0.901,
+      "step": 7541
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6957437390941274e-06,
+      "loss": 0.8537,
+      "step": 7542
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6933252193480585e-06,
+      "loss": 0.8946,
+      "step": 7543
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6909073119619555e-06,
+      "loss": 0.8457,
+      "step": 7544
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.688490017170595e-06,
+      "loss": 0.8548,
+      "step": 7545
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6860733352086866e-06,
+      "loss": 0.8038,
+      "step": 7546
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6836572663108803e-06,
+      "loss": 0.9148,
+      "step": 7547
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6812418107117765e-06,
+      "loss": 0.8946,
+      "step": 7548
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6788269686459044e-06,
+      "loss": 0.7825,
+      "step": 7549
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6764127403477347e-06,
+      "loss": 0.8786,
+      "step": 7550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6739991260516905e-06,
+      "loss": 0.8189,
+      "step": 7551
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6715861259921226e-06,
+      "loss": 0.9155,
+      "step": 7552
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6691737404033223e-06,
+      "loss": 0.8508,
+      "step": 7553
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6667619695195287e-06,
+      "loss": 0.8329,
+      "step": 7554
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.664350813574923e-06,
+      "loss": 0.8786,
+      "step": 7555
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6619402728036157e-06,
+      "loss": 0.8599,
+      "step": 7556
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6595303474396617e-06,
+      "loss": 0.8298,
+      "step": 7557
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.657121037717064e-06,
+      "loss": 0.7855,
+      "step": 7558
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.654712343869756e-06,
+      "loss": 0.8651,
+      "step": 7559
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.652304266131612e-06,
+      "loss": 0.8589,
+      "step": 7560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.649896804736456e-06,
+      "loss": 0.812,
+      "step": 7561
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6474899599180426e-06,
+      "loss": 0.901,
+      "step": 7562
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6450837319100653e-06,
+      "loss": 0.8242,
+      "step": 7563
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.642678120946168e-06,
+      "loss": 0.9122,
+      "step": 7564
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6402731272599245e-06,
+      "loss": 0.8808,
+      "step": 7565
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6378687510848576e-06,
+      "loss": 0.8291,
+      "step": 7566
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6354649926544183e-06,
+      "loss": 0.793,
+      "step": 7567
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6330618522020124e-06,
+      "loss": 0.887,
+      "step": 7568
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6306593299609737e-06,
+      "loss": 0.8913,
+      "step": 7569
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6282574261645776e-06,
+      "loss": 0.8519,
+      "step": 7570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.625856141046048e-06,
+      "loss": 0.8769,
+      "step": 7571
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.62345547483854e-06,
+      "loss": 0.8801,
+      "step": 7572
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6210554277751475e-06,
+      "loss": 0.8239,
+      "step": 7573
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.618656000088916e-06,
+      "loss": 0.8411,
+      "step": 7574
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6162571920128187e-06,
+      "loss": 0.8238,
+      "step": 7575
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6138590037797695e-06,
+      "loss": 0.8815,
+      "step": 7576
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.61146143562263e-06,
+      "loss": 0.8928,
+      "step": 7577
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6090644877741986e-06,
+      "loss": 0.8599,
+      "step": 7578
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6066681604672105e-06,
+      "loss": 0.9472,
+      "step": 7579
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6042724539343378e-06,
+      "loss": 0.8348,
+      "step": 7580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6018773684082032e-06,
+      "loss": 0.8933,
+      "step": 7581
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.599482904121361e-06,
+      "loss": 0.8611,
+      "step": 7582
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5970890613063014e-06,
+      "loss": 0.8944,
+      "step": 7583
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.594695840195468e-06,
+      "loss": 0.8231,
+      "step": 7584
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.592303241021231e-06,
+      "loss": 0.8711,
+      "step": 7585
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5899112640159017e-06,
+      "loss": 0.8785,
+      "step": 7586
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.587519909411742e-06,
+      "loss": 0.9376,
+      "step": 7587
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.585129177440938e-06,
+      "loss": 0.8851,
+      "step": 7588
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5827390683356257e-06,
+      "loss": 0.9117,
+      "step": 7589
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.580349582327882e-06,
+      "loss": 0.8973,
+      "step": 7590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.577960719649717e-06,
+      "loss": 0.9106,
+      "step": 7591
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.575572480533076e-06,
+      "loss": 0.8418,
+      "step": 7592
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5731848652098588e-06,
+      "loss": 0.8761,
+      "step": 7593
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.570797873911892e-06,
+      "loss": 0.7874,
+      "step": 7594
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5684115068709434e-06,
+      "loss": 0.8585,
+      "step": 7595
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.566025764318728e-06,
+      "loss": 0.8514,
+      "step": 7596
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.563640646486891e-06,
+      "loss": 0.8426,
+      "step": 7597
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5612561536070213e-06,
+      "loss": 0.8297,
+      "step": 7598
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.558872285910642e-06,
+      "loss": 0.8142,
+      "step": 7599
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5564890436292243e-06,
+      "loss": 0.8543,
+      "step": 7600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.554106426994176e-06,
+      "loss": 0.8531,
+      "step": 7601
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5517244362368363e-06,
+      "loss": 0.8753,
+      "step": 7602
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.549343071588498e-06,
+      "loss": 0.8646,
+      "step": 7603
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5469623332803795e-06,
+      "loss": 0.8962,
+      "step": 7604
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.54458222154364e-06,
+      "loss": 0.8471,
+      "step": 7605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5422027366093893e-06,
+      "loss": 0.8289,
+      "step": 7606
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5398238787086646e-06,
+      "loss": 0.794,
+      "step": 7607
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5374456480724427e-06,
+      "loss": 0.8551,
+      "step": 7608
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.53506804493165e-06,
+      "loss": 0.9097,
+      "step": 7609
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.532691069517142e-06,
+      "loss": 0.8318,
+      "step": 7610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.530314722059712e-06,
+      "loss": 0.8022,
+      "step": 7611
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5279390027901004e-06,
+      "loss": 0.8218,
+      "step": 7612
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5255639119389862e-06,
+      "loss": 0.869,
+      "step": 7613
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5231894497369802e-06,
+      "loss": 0.9062,
+      "step": 7614
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5208156164146325e-06,
+      "loss": 0.9421,
+      "step": 7615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5184424122024406e-06,
+      "loss": 0.8331,
+      "step": 7616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5160698373308357e-06,
+      "loss": 0.8692,
+      "step": 7617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5136978920301822e-06,
+      "loss": 0.901,
+      "step": 7618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.511326576530797e-06,
+      "loss": 0.867,
+      "step": 7619
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.508955891062924e-06,
+      "loss": 0.892,
+      "step": 7620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.506585835856747e-06,
+      "loss": 0.8773,
+      "step": 7621
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5042164111423983e-06,
+      "loss": 0.8119,
+      "step": 7622
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.501847617149935e-06,
+      "loss": 0.8727,
+      "step": 7623
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4994794541093667e-06,
+      "loss": 0.8445,
+      "step": 7624
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.49711192225063e-06,
+      "loss": 0.8612,
+      "step": 7625
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4947450218036106e-06,
+      "loss": 0.8208,
+      "step": 7626
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4923787529981256e-06,
+      "loss": 0.8617,
+      "step": 7627
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4900131160639283e-06,
+      "loss": 0.8268,
+      "step": 7628
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4876481112307225e-06,
+      "loss": 0.8981,
+      "step": 7629
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.485283738728139e-06,
+      "loss": 0.8633,
+      "step": 7630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.482919998785751e-06,
+      "loss": 0.7245,
+      "step": 7631
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4805568916330747e-06,
+      "loss": 0.8835,
+      "step": 7632
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4781944174995584e-06,
+      "loss": 0.8799,
+      "step": 7633
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4758325766145896e-06,
+      "loss": 0.8464,
+      "step": 7634
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4734713692074973e-06,
+      "loss": 0.9332,
+      "step": 7635
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.471110795507554e-06,
+      "loss": 0.873,
+      "step": 7636
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4687508557439586e-06,
+      "loss": 0.8498,
+      "step": 7637
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4663915501458523e-06,
+      "loss": 0.9108,
+      "step": 7638
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.464032878942324e-06,
+      "loss": 0.9143,
+      "step": 7639
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4616748423623893e-06,
+      "loss": 0.884,
+      "step": 7640
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.459317440635005e-06,
+      "loss": 0.9137,
+      "step": 7641
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4569606739890737e-06,
+      "loss": 0.8964,
+      "step": 7642
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4546045426534267e-06,
+      "loss": 0.8641,
+      "step": 7643
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.452249046856836e-06,
+      "loss": 0.8881,
+      "step": 7644
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.449894186828019e-06,
+      "loss": 0.921,
+      "step": 7645
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4475399627956197e-06,
+      "loss": 0.924,
+      "step": 7646
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4451863749882287e-06,
+      "loss": 0.8559,
+      "step": 7647
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4428334236343774e-06,
+      "loss": 0.9299,
+      "step": 7648
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4404811089625267e-06,
+      "loss": 0.8565,
+      "step": 7649
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.438129431201075e-06,
+      "loss": 0.8213,
+      "step": 7650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4357783905783727e-06,
+      "loss": 0.8659,
+      "step": 7651
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.433427987322693e-06,
+      "loss": 0.8035,
+      "step": 7652
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4310782216622516e-06,
+      "loss": 0.8839,
+      "step": 7653
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4287290938252103e-06,
+      "loss": 0.9209,
+      "step": 7654
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4263806040396596e-06,
+      "loss": 0.8344,
+      "step": 7655
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.424032752533627e-06,
+      "loss": 0.7571,
+      "step": 7656
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.421685539535089e-06,
+      "loss": 0.8583,
+      "step": 7657
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4193389652719478e-06,
+      "loss": 0.8831,
+      "step": 7658
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4169930299720542e-06,
+      "loss": 0.8325,
+      "step": 7659
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4146477338631856e-06,
+      "loss": 0.8186,
+      "step": 7660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4123030771730704e-06,
+      "loss": 0.7837,
+      "step": 7661
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4099590601293632e-06,
+      "loss": 0.8274,
+      "step": 7662
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.407615682959661e-06,
+      "loss": 0.8873,
+      "step": 7663
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4052729458915024e-06,
+      "loss": 0.8663,
+      "step": 7664
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4029308491523593e-06,
+      "loss": 0.8603,
+      "step": 7665
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4005893929696377e-06,
+      "loss": 0.8685,
+      "step": 7666
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3982485775706942e-06,
+      "loss": 0.8003,
+      "step": 7667
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3959084031828114e-06,
+      "loss": 0.8676,
+      "step": 7668
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3935688700332104e-06,
+      "loss": 0.8908,
+      "step": 7669
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3912299783490567e-06,
+      "loss": 0.8042,
+      "step": 7670
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3888917283574516e-06,
+      "loss": 0.8539,
+      "step": 7671
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3865541202854314e-06,
+      "loss": 0.8091,
+      "step": 7672
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3842171543599656e-06,
+      "loss": 0.8539,
+      "step": 7673
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3818808308079753e-06,
+      "loss": 0.7493,
+      "step": 7674
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.379545149856306e-06,
+      "loss": 0.9066,
+      "step": 7675
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3772101117317437e-06,
+      "loss": 0.8433,
+      "step": 7676
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3748757166610204e-06,
+      "loss": 0.8144,
+      "step": 7677
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.372541964870795e-06,
+      "loss": 0.8263,
+      "step": 7678
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3702088565876646e-06,
+      "loss": 0.8701,
+      "step": 7679
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.367876392038174e-06,
+      "loss": 0.8095,
+      "step": 7680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3655445714487934e-06,
+      "loss": 0.8997,
+      "step": 7681
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.363213395045941e-06,
+      "loss": 0.8705,
+      "step": 7682
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.360882863055962e-06,
+      "loss": 0.8063,
+      "step": 7683
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3585529757051504e-06,
+      "loss": 0.8572,
+      "step": 7684
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3562237332197277e-06,
+      "loss": 0.8605,
+      "step": 7685
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.353895135825854e-06,
+      "loss": 0.8928,
+      "step": 7686
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3515671837496355e-06,
+      "loss": 0.8429,
+      "step": 7687
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3492398772171074e-06,
+      "loss": 0.8491,
+      "step": 7688
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3469132164542394e-06,
+      "loss": 0.8265,
+      "step": 7689
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.344587201686952e-06,
+      "loss": 0.8311,
+      "step": 7690
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3422618331410904e-06,
+      "loss": 0.8098,
+      "step": 7691
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3399371110424372e-06,
+      "loss": 0.8195,
+      "step": 7692
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3376130356167213e-06,
+      "loss": 0.8555,
+      "step": 7693
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3352896070896057e-06,
+      "loss": 0.8503,
+      "step": 7694
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3329668256866853e-06,
+      "loss": 0.8765,
+      "step": 7695
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.330644691633492e-06,
+      "loss": 0.8657,
+      "step": 7696
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3283232051555057e-06,
+      "loss": 0.8039,
+      "step": 7697
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3260023664781326e-06,
+      "loss": 0.7349,
+      "step": 7698
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3236821758267167e-06,
+      "loss": 0.834,
+      "step": 7699
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.321362633426547e-06,
+      "loss": 0.8254,
+      "step": 7700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.319043739502842e-06,
+      "loss": 0.8626,
+      "step": 7701
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.316725494280757e-06,
+      "loss": 0.9197,
+      "step": 7702
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3144078979853913e-06,
+      "loss": 0.8855,
+      "step": 7703
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3120909508417754e-06,
+      "loss": 0.845,
+      "step": 7704
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.309774653074872e-06,
+      "loss": 0.87,
+      "step": 7705
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.307459004909599e-06,
+      "loss": 0.811,
+      "step": 7706
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.305144006570792e-06,
+      "loss": 0.8392,
+      "step": 7707
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3028296582832285e-06,
+      "loss": 0.8776,
+      "step": 7708
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.300515960271632e-06,
+      "loss": 0.8768,
+      "step": 7709
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2982029127606517e-06,
+      "loss": 0.9036,
+      "step": 7710
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2958905159748744e-06,
+      "loss": 0.9132,
+      "step": 7711
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2935787701388346e-06,
+      "loss": 0.9083,
+      "step": 7712
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2912676754769925e-06,
+      "loss": 0.8813,
+      "step": 7713
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2889572322137454e-06,
+      "loss": 0.8328,
+      "step": 7714
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2866474405734385e-06,
+      "loss": 0.8447,
+      "step": 7715
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2843383007803364e-06,
+      "loss": 0.8083,
+      "step": 7716
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2820298130586593e-06,
+      "loss": 0.8763,
+      "step": 7717
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.279721977632546e-06,
+      "loss": 0.8583,
+      "step": 7718
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.277414794726089e-06,
+      "loss": 0.7988,
+      "step": 7719
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.275108264563306e-06,
+      "loss": 0.8699,
+      "step": 7720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.272802387368149e-06,
+      "loss": 0.8935,
+      "step": 7721
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.270497163364521e-06,
+      "loss": 0.8298,
+      "step": 7722
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2681925927762482e-06,
+      "loss": 0.9142,
+      "step": 7723
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2658886758270947e-06,
+      "loss": 0.9108,
+      "step": 7724
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2635854127407697e-06,
+      "loss": 0.8785,
+      "step": 7725
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2612828037409116e-06,
+      "loss": 0.9038,
+      "step": 7726
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2589808490510934e-06,
+      "loss": 0.831,
+      "step": 7727
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.256679548894831e-06,
+      "loss": 0.8225,
+      "step": 7728
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2543789034955786e-06,
+      "loss": 0.8282,
+      "step": 7729
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.252078913076718e-06,
+      "loss": 0.8736,
+      "step": 7730
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.249779577861568e-06,
+      "loss": 0.8739,
+      "step": 7731
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.247480898073395e-06,
+      "loss": 0.8647,
+      "step": 7732
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.24518287393539e-06,
+      "loss": 0.8292,
+      "step": 7733
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.242885505670681e-06,
+      "loss": 0.8234,
+      "step": 7734
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2405887935023437e-06,
+      "loss": 0.872,
+      "step": 7735
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.238292737653379e-06,
+      "loss": 0.8351,
+      "step": 7736
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2359973383467227e-06,
+      "loss": 0.852,
+      "step": 7737
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.233702595805258e-06,
+      "loss": 0.8547,
+      "step": 7738
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.231408510251797e-06,
+      "loss": 0.8607,
+      "step": 7739
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.229115081909082e-06,
+      "loss": 0.9181,
+      "step": 7740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2268223109998042e-06,
+      "loss": 0.8433,
+      "step": 7741
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.224530197746587e-06,
+      "loss": 0.9263,
+      "step": 7742
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2222387423719847e-06,
+      "loss": 0.8238,
+      "step": 7743
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2199479450984892e-06,
+      "loss": 0.8359,
+      "step": 7744
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2176578061485354e-06,
+      "loss": 0.9098,
+      "step": 7745
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2153683257444856e-06,
+      "loss": 0.7811,
+      "step": 7746
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2130795041086404e-06,
+      "loss": 0.7917,
+      "step": 7747
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.210791341463243e-06,
+      "loss": 0.8698,
+      "step": 7748
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2085038380304636e-06,
+      "loss": 0.8962,
+      "step": 7749
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.206216994032411e-06,
+      "loss": 0.8936,
+      "step": 7750
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2039308096911317e-06,
+      "loss": 0.9419,
+      "step": 7751
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2016452852286127e-06,
+      "loss": 0.8669,
+      "step": 7752
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.199360420866768e-06,
+      "loss": 0.8234,
+      "step": 7753
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1970762168274495e-06,
+      "loss": 0.8777,
+      "step": 7754
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.194792673332451e-06,
+      "loss": 0.8199,
+      "step": 7755
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1925097906034962e-06,
+      "loss": 0.7394,
+      "step": 7756
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.190227568862243e-06,
+      "loss": 0.8676,
+      "step": 7757
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.187946008330295e-06,
+      "loss": 0.9095,
+      "step": 7758
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.185665109229181e-06,
+      "loss": 0.8492,
+      "step": 7759
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1833848717803674e-06,
+      "loss": 0.8152,
+      "step": 7760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.181105296205266e-06,
+      "loss": 0.8727,
+      "step": 7761
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.178826382725212e-06,
+      "loss": 0.8083,
+      "step": 7762
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1765481315614767e-06,
+      "loss": 0.8251,
+      "step": 7763
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1742705429352827e-06,
+      "loss": 0.8179,
+      "step": 7764
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1719936170677723e-06,
+      "loss": 0.8498,
+      "step": 7765
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1697173541800254e-06,
+      "loss": 0.8619,
+      "step": 7766
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.167441754493066e-06,
+      "loss": 0.8035,
+      "step": 7767
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.165166818227845e-06,
+      "loss": 0.8483,
+      "step": 7768
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1628925456052493e-06,
+      "loss": 0.8083,
+      "step": 7769
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1606189368461117e-06,
+      "loss": 0.8708,
+      "step": 7770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1583459921711878e-06,
+      "loss": 0.9052,
+      "step": 7771
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.156073711801172e-06,
+      "loss": 0.8301,
+      "step": 7772
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.153802095956703e-06,
+      "loss": 0.8082,
+      "step": 7773
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.151531144858344e-06,
+      "loss": 0.86,
+      "step": 7774
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.149260858726596e-06,
+      "loss": 0.8141,
+      "step": 7775
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.146991237781899e-06,
+      "loss": 0.8192,
+      "step": 7776
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1447222822446298e-06,
+      "loss": 0.8596,
+      "step": 7777
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.142453992335096e-06,
+      "loss": 0.7961,
+      "step": 7778
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1401863682735367e-06,
+      "loss": 0.8304,
+      "step": 7779
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.137919410280139e-06,
+      "loss": 0.8325,
+      "step": 7780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.135653118575015e-06,
+      "loss": 0.8635,
+      "step": 7781
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1333874933782114e-06,
+      "loss": 0.8475,
+      "step": 7782
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1311225349097208e-06,
+      "loss": 0.9223,
+      "step": 7783
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.128858243389461e-06,
+      "loss": 0.8086,
+      "step": 7784
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.126594619037284e-06,
+      "loss": 0.9246,
+      "step": 7785
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.124331662072987e-06,
+      "loss": 0.9371,
+      "step": 7786
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1220693727162964e-06,
+      "loss": 0.8745,
+      "step": 7787
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.119807751186872e-06,
+      "loss": 0.8797,
+      "step": 7788
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1175467977043084e-06,
+      "loss": 0.8298,
+      "step": 7789
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1152865124881436e-06,
+      "loss": 0.791,
+      "step": 7790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.113026895757841e-06,
+      "loss": 0.917,
+      "step": 7791
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.110767947732801e-06,
+      "loss": 0.9171,
+      "step": 7792
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1085096686323667e-06,
+      "loss": 0.7829,
+      "step": 7793
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.106252058675806e-06,
+      "loss": 0.9003,
+      "step": 7794
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.103995118082326e-06,
+      "loss": 0.8268,
+      "step": 7795
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.101738847071072e-06,
+      "loss": 0.8609,
+      "step": 7796
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0994832458611215e-06,
+      "loss": 0.8101,
+      "step": 7797
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.097228314671481e-06,
+      "loss": 0.8425,
+      "step": 7798
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0949740537211027e-06,
+      "loss": 0.7479,
+      "step": 7799
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.092720463228872e-06,
+      "loss": 0.8381,
+      "step": 7800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0904675434136024e-06,
+      "loss": 0.8483,
+      "step": 7801
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0882152944940423e-06,
+      "loss": 0.8214,
+      "step": 7802
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.085963716688886e-06,
+      "loss": 0.7635,
+      "step": 7803
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0837128102167514e-06,
+      "loss": 0.8489,
+      "step": 7804
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0814625752961925e-06,
+      "loss": 0.8441,
+      "step": 7805
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.079213012145705e-06,
+      "loss": 0.908,
+      "step": 7806
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0769641209837142e-06,
+      "loss": 0.8151,
+      "step": 7807
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0747159020285766e-06,
+      "loss": 0.7794,
+      "step": 7808
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0724683554985944e-06,
+      "loss": 0.8423,
+      "step": 7809
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0702214816119925e-06,
+      "loss": 0.7968,
+      "step": 7810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.067975280586937e-06,
+      "loss": 0.8091,
+      "step": 7811
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.065729752641532e-06,
+      "loss": 0.8589,
+      "step": 7812
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0634848979938082e-06,
+      "loss": 0.8774,
+      "step": 7813
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0612407168617352e-06,
+      "loss": 0.8639,
+      "step": 7814
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0589972094632137e-06,
+      "loss": 0.8858,
+      "step": 7815
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0567543760160866e-06,
+      "loss": 0.843,
+      "step": 7816
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0545122167381236e-06,
+      "loss": 0.9495,
+      "step": 7817
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.05227073184703e-06,
+      "loss": 0.8258,
+      "step": 7818
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0500299215604534e-06,
+      "loss": 0.8216,
+      "step": 7819
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.047789786095967e-06,
+      "loss": 0.8675,
+      "step": 7820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.045550325671074e-06,
+      "loss": 0.9232,
+      "step": 7821
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0433115405032334e-06,
+      "loss": 0.9061,
+      "step": 7822
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0410734308098177e-06,
+      "loss": 0.8035,
+      "step": 7823
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0388359968081395e-06,
+      "loss": 0.8078,
+      "step": 7824
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.036599238715451e-06,
+      "loss": 0.8884,
+      "step": 7825
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.034363156748933e-06,
+      "loss": 0.8721,
+      "step": 7826
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0321277511257e-06,
+      "loss": 0.8935,
+      "step": 7827
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0298930220628086e-06,
+      "loss": 0.8672,
+      "step": 7828
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.027658969777243e-06,
+      "loss": 0.8944,
+      "step": 7829
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.025425594485919e-06,
+      "loss": 0.8221,
+      "step": 7830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0231928964056968e-06,
+      "loss": 0.8679,
+      "step": 7831
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0209608757533626e-06,
+      "loss": 0.8817,
+      "step": 7832
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0187295327456355e-06,
+      "loss": 0.9,
+      "step": 7833
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0164988675991768e-06,
+      "loss": 0.8685,
+      "step": 7834
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0142688805305796e-06,
+      "loss": 0.8524,
+      "step": 7835
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0120395717563655e-06,
+      "loss": 0.8184,
+      "step": 7836
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0098109414929932e-06,
+      "loss": 0.8455,
+      "step": 7837
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.00758298995686e-06,
+      "loss": 0.8295,
+      "step": 7838
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.005355717364292e-06,
+      "loss": 0.9417,
+      "step": 7839
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0031291239315473e-06,
+      "loss": 0.8792,
+      "step": 7840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.000903209874828e-06,
+      "loss": 0.9248,
+      "step": 7841
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9986779754102613e-06,
+      "loss": 0.8702,
+      "step": 7842
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9964534207539064e-06,
+      "loss": 0.8397,
+      "step": 7843
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9942295461217698e-06,
+      "loss": 0.7735,
+      "step": 7844
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.992006351729775e-06,
+      "loss": 0.8803,
+      "step": 7845
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9897838377937947e-06,
+      "loss": 0.803,
+      "step": 7846
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9875620045296237e-06,
+      "loss": 0.8324,
+      "step": 7847
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.985340852152999e-06,
+      "loss": 0.8649,
+      "step": 7848
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9831203808795883e-06,
+      "loss": 0.8915,
+      "step": 7849
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9809005909249866e-06,
+      "loss": 0.7963,
+      "step": 7850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9786814825047384e-06,
+      "loss": 0.7563,
+      "step": 7851
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9764630558343064e-06,
+      "loss": 0.8346,
+      "step": 7852
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9742453111290937e-06,
+      "loss": 0.8245,
+      "step": 7853
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9720282486044407e-06,
+      "loss": 0.947,
+      "step": 7854
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.969811868475615e-06,
+      "loss": 0.928,
+      "step": 7855
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9675961709578194e-06,
+      "loss": 0.8654,
+      "step": 7856
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9653811562661927e-06,
+      "loss": 0.8485,
+      "step": 7857
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9631668246158105e-06,
+      "loss": 0.8713,
+      "step": 7858
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.960953176221675e-06,
+      "loss": 0.8017,
+      "step": 7859
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.958740211298722e-06,
+      "loss": 0.8802,
+      "step": 7860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.95652793006183e-06,
+      "loss": 0.923,
+      "step": 7861
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9543163327258016e-06,
+      "loss": 0.8671,
+      "step": 7862
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9521054195053745e-06,
+      "loss": 0.9354,
+      "step": 7863
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.949895190615227e-06,
+      "loss": 0.8498,
+      "step": 7864
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.947685646269964e-06,
+      "loss": 0.8601,
+      "step": 7865
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9454767866841225e-06,
+      "loss": 0.8009,
+      "step": 7866
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.943268612072183e-06,
+      "loss": 0.9055,
+      "step": 7867
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.941061122648545e-06,
+      "loss": 0.9066,
+      "step": 7868
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.938854318627554e-06,
+      "loss": 0.8517,
+      "step": 7869
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9366482002234874e-06,
+      "loss": 0.8475,
+      "step": 7870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9344427676505493e-06,
+      "loss": 0.7983,
+      "step": 7871
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.932238021122877e-06,
+      "loss": 0.8349,
+      "step": 7872
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9300339608545536e-06,
+      "loss": 0.8436,
+      "step": 7873
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9278305870595814e-06,
+      "loss": 0.8618,
+      "step": 7874
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.925627899951903e-06,
+      "loss": 0.8743,
+      "step": 7875
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.92342589974539e-06,
+      "loss": 0.8889,
+      "step": 7876
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9212245866538558e-06,
+      "loss": 0.7916,
+      "step": 7877
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.919023960891039e-06,
+      "loss": 0.8674,
+      "step": 7878
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9168240226706102e-06,
+      "loss": 0.757,
+      "step": 7879
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9146247722061806e-06,
+      "loss": 0.8528,
+      "step": 7880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.912426209711294e-06,
+      "loss": 0.8558,
+      "step": 7881
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.910228335399419e-06,
+      "loss": 0.8571,
+      "step": 7882
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9080311494839685e-06,
+      "loss": 0.8028,
+      "step": 7883
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.90583465217828e-06,
+      "loss": 0.8695,
+      "step": 7884
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.903638843695624e-06,
+      "loss": 0.8575,
+      "step": 7885
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9014437242492133e-06,
+      "loss": 0.8018,
+      "step": 7886
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8992492940521844e-06,
+      "loss": 0.8564,
+      "step": 7887
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.897055553317607e-06,
+      "loss": 0.8262,
+      "step": 7888
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.894862502258493e-06,
+      "loss": 0.8241,
+      "step": 7889
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.89267014108778e-06,
+      "loss": 0.9384,
+      "step": 7890
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.890478470018334e-06,
+      "loss": 0.783,
+      "step": 7891
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8882874892629654e-06,
+      "loss": 0.8534,
+      "step": 7892
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8860971990344144e-06,
+      "loss": 0.791,
+      "step": 7893
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.883907599545348e-06,
+      "loss": 0.8582,
+      "step": 7894
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8817186910083683e-06,
+      "loss": 0.9044,
+      "step": 7895
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8795304736360184e-06,
+      "loss": 0.8834,
+      "step": 7896
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.877342947640763e-06,
+      "loss": 0.8698,
+      "step": 7897
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8751561132350025e-06,
+      "loss": 0.8909,
+      "step": 7898
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.872969970631079e-06,
+      "loss": 0.8716,
+      "step": 7899
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8707845200412567e-06,
+      "loss": 0.8972,
+      "step": 7900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.868599761677733e-06,
+      "loss": 0.8965,
+      "step": 7901
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.866415695752649e-06,
+      "loss": 0.9324,
+      "step": 7902
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8642323224780656e-06,
+      "loss": 0.8647,
+      "step": 7903
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.862049642065986e-06,
+      "loss": 0.872,
+      "step": 7904
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.859867654728338e-06,
+      "loss": 0.865,
+      "step": 7905
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.857686360676991e-06,
+      "loss": 0.8317,
+      "step": 7906
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8555057601237402e-06,
+      "loss": 0.8232,
+      "step": 7907
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.853325853280312e-06,
+      "loss": 0.9414,
+      "step": 7908
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8511466403583766e-06,
+      "loss": 0.8197,
+      "step": 7909
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8489681215695242e-06,
+      "loss": 0.8741,
+      "step": 7910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8467902971252815e-06,
+      "loss": 0.8227,
+      "step": 7911
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8446131672371136e-06,
+      "loss": 0.8201,
+      "step": 7912
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8424367321164115e-06,
+      "loss": 0.8294,
+      "step": 7913
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.840260991974497e-06,
+      "loss": 0.8285,
+      "step": 7914
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.838085947022633e-06,
+      "loss": 0.7987,
+      "step": 7915
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.83591159747201e-06,
+      "loss": 0.8103,
+      "step": 7916
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.833737943533751e-06,
+      "loss": 0.8214,
+      "step": 7917
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8315649854189066e-06,
+      "loss": 0.8413,
+      "step": 7918
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.829392723338472e-06,
+      "loss": 0.9518,
+      "step": 7919
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8272211575033635e-06,
+      "loss": 0.8507,
+      "step": 7920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8250502881244325e-06,
+      "loss": 0.8742,
+      "step": 7921
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8228801154124687e-06,
+      "loss": 0.8539,
+      "step": 7922
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.820710639578187e-06,
+      "loss": 0.8643,
+      "step": 7923
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8185418608322344e-06,
+      "loss": 0.876,
+      "step": 7924
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8163737793851998e-06,
+      "loss": 0.8745,
+      "step": 7925
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.814206395447593e-06,
+      "loss": 0.8332,
+      "step": 7926
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8120397092298567e-06,
+      "loss": 0.8886,
+      "step": 7927
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.80987372094238e-06,
+      "loss": 0.9034,
+      "step": 7928
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8077084307954695e-06,
+      "loss": 0.7966,
+      "step": 7929
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.805543838999364e-06,
+      "loss": 0.8448,
+      "step": 7930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8033799457642475e-06,
+      "loss": 0.8093,
+      "step": 7931
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.801216751300223e-06,
+      "loss": 0.8737,
+      "step": 7932
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7990542558173286e-06,
+      "loss": 0.8597,
+      "step": 7933
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.796892459525541e-06,
+      "loss": 0.8531,
+      "step": 7934
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.794731362634763e-06,
+      "loss": 0.8629,
+      "step": 7935
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7925709653548295e-06,
+      "loss": 0.8173,
+      "step": 7936
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7904112678955064e-06,
+      "loss": 0.8877,
+      "step": 7937
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.788252270466497e-06,
+      "loss": 0.8708,
+      "step": 7938
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.786093973277437e-06,
+      "loss": 0.8246,
+      "step": 7939
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.783936376537886e-06,
+      "loss": 0.9185,
+      "step": 7940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.781779480457344e-06,
+      "loss": 0.8868,
+      "step": 7941
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7796232852452378e-06,
+      "loss": 0.8951,
+      "step": 7942
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7774677911109247e-06,
+      "loss": 0.8248,
+      "step": 7943
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.775312998263703e-06,
+      "loss": 0.8817,
+      "step": 7944
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.773158906912793e-06,
+      "loss": 0.8333,
+      "step": 7945
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.771005517267349e-06,
+      "loss": 0.8878,
+      "step": 7946
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7688528295364634e-06,
+      "loss": 0.8596,
+      "step": 7947
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7667008439291552e-06,
+      "loss": 0.8072,
+      "step": 7948
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7645495606543704e-06,
+      "loss": 0.84,
+      "step": 7949
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.762398979920998e-06,
+      "loss": 0.8528,
+      "step": 7950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.760249101937855e-06,
+      "loss": 0.8644,
+      "step": 7951
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7580999269136854e-06,
+      "loss": 0.8228,
+      "step": 7952
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7559514550571643e-06,
+      "loss": 0.884,
+      "step": 7953
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7538036865769093e-06,
+      "loss": 0.8401,
+      "step": 7954
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.751656621681458e-06,
+      "loss": 0.8612,
+      "step": 7955
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.749510260579282e-06,
+      "loss": 0.8292,
+      "step": 7956
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7473646034787936e-06,
+      "loss": 0.8275,
+      "step": 7957
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7452196505883265e-06,
+      "loss": 0.8994,
+      "step": 7958
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7430754021161464e-06,
+      "loss": 0.7914,
+      "step": 7959
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7409318582704594e-06,
+      "loss": 0.9194,
+      "step": 7960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7387890192593937e-06,
+      "loss": 0.8977,
+      "step": 7961
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.736646885291011e-06,
+      "loss": 0.8321,
+      "step": 7962
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7345054565733097e-06,
+      "loss": 0.8307,
+      "step": 7963
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7323647333142176e-06,
+      "loss": 0.8465,
+      "step": 7964
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7302247157215913e-06,
+      "loss": 0.8726,
+      "step": 7965
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.728085404003217e-06,
+      "loss": 0.8184,
+      "step": 7966
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.725946798366821e-06,
+      "loss": 0.8151,
+      "step": 7967
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7238088990200538e-06,
+      "loss": 0.7981,
+      "step": 7968
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7216717061704956e-06,
+      "loss": 0.9106,
+      "step": 7969
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7195352200256675e-06,
+      "loss": 0.8609,
+      "step": 7970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.717399440793014e-06,
+      "loss": 0.8665,
+      "step": 7971
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7152643686799095e-06,
+      "loss": 0.8882,
+      "step": 7972
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.713130003893666e-06,
+      "loss": 0.8036,
+      "step": 7973
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.710996346641528e-06,
+      "loss": 0.8705,
+      "step": 7974
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7088633971306643e-06,
+      "loss": 0.8216,
+      "step": 7975
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.706731155568175e-06,
+      "loss": 0.7793,
+      "step": 7976
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7045996221611006e-06,
+      "loss": 0.8271,
+      "step": 7977
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7024687971164032e-06,
+      "loss": 0.8305,
+      "step": 7978
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.700338680640978e-06,
+      "loss": 0.8234,
+      "step": 7979
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.698209272941659e-06,
+      "loss": 0.8295,
+      "step": 7980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6960805742252014e-06,
+      "loss": 0.86,
+      "step": 7981
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.693952584698294e-06,
+      "loss": 0.919,
+      "step": 7982
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6918253045675634e-06,
+      "loss": 0.8121,
+      "step": 7983
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.689698734039561e-06,
+      "loss": 0.8611,
+      "step": 7984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.687572873320763e-06,
+      "loss": 0.7928,
+      "step": 7985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.685447722617597e-06,
+      "loss": 0.8082,
+      "step": 7986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.683323282136403e-06,
+      "loss": 0.8707,
+      "step": 7987
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6811995520834543e-06,
+      "loss": 0.8684,
+      "step": 7988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.679076532664967e-06,
+      "loss": 0.8681,
+      "step": 7989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.676954224087075e-06,
+      "loss": 0.8223,
+      "step": 7990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6748326265558465e-06,
+      "loss": 0.8857,
+      "step": 7991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6727117402772886e-06,
+      "loss": 0.9389,
+      "step": 7992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.670591565457329e-06,
+      "loss": 0.8348,
+      "step": 7993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.668472102301829e-06,
+      "loss": 0.8527,
+      "step": 7994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6663533510165886e-06,
+      "loss": 0.8818,
+      "step": 7995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.664235311807327e-06,
+      "loss": 0.8634,
+      "step": 7996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6621179848796995e-06,
+      "loss": 0.8752,
+      "step": 7997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6600013704392946e-06,
+      "loss": 0.8179,
+      "step": 7998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6578854686916323e-06,
+      "loss": 0.9123,
+      "step": 7999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.655770279842157e-06,
+      "loss": 0.8691,
+      "step": 8000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6536558040962444e-06,
+      "loss": 0.8547,
+      "step": 8001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.651542041659211e-06,
+      "loss": 0.8568,
+      "step": 8002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6494289927362937e-06,
+      "loss": 0.8536,
+      "step": 8003
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6473166575326603e-06,
+      "loss": 0.8527,
+      "step": 8004
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.645205036253419e-06,
+      "loss": 0.7919,
+      "step": 8005
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6430941291035984e-06,
+      "loss": 0.8528,
+      "step": 8006
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6409839362881597e-06,
+      "loss": 0.844,
+      "step": 8007
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6388744580119975e-06,
+      "loss": 0.8633,
+      "step": 8008
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.63676569447994e-06,
+      "loss": 0.8412,
+      "step": 8009
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6346576458967397e-06,
+      "loss": 0.8337,
+      "step": 8010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.632550312467079e-06,
+      "loss": 0.8453,
+      "step": 8011
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.630443694395579e-06,
+      "loss": 0.8645,
+      "step": 8012
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.628337791886785e-06,
+      "loss": 0.8207,
+      "step": 8013
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6262326051451683e-06,
+      "loss": 0.7864,
+      "step": 8014
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6241281343751447e-06,
+      "loss": 0.7957,
+      "step": 8015
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6220243797810483e-06,
+      "loss": 0.818,
+      "step": 8016
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6199213415671452e-06,
+      "loss": 0.8217,
+      "step": 8017
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6178190199376394e-06,
+      "loss": 0.8946,
+      "step": 8018
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.615717415096659e-06,
+      "loss": 0.7892,
+      "step": 8019
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6136165272482596e-06,
+      "loss": 0.809,
+      "step": 8020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.611516356596434e-06,
+      "loss": 0.8277,
+      "step": 8021
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6094169033451066e-06,
+      "loss": 0.8975,
+      "step": 8022
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6073181676981253e-06,
+      "loss": 0.869,
+      "step": 8023
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.605220149859267e-06,
+      "loss": 0.8479,
+      "step": 8024
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.60312285003225e-06,
+      "loss": 0.8825,
+      "step": 8025
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6010262684207134e-06,
+      "loss": 0.83,
+      "step": 8026
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.598930405228226e-06,
+      "loss": 0.855,
+      "step": 8027
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.596835260658297e-06,
+      "loss": 0.8149,
+      "step": 8028
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5947408349143534e-06,
+      "loss": 0.8409,
+      "step": 8029
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5926471281997577e-06,
+      "loss": 0.8599,
+      "step": 8030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.590554140717807e-06,
+      "loss": 0.8534,
+      "step": 8031
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.588461872671719e-06,
+      "loss": 0.8828,
+      "step": 8032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.586370324264653e-06,
+      "loss": 0.8613,
+      "step": 8033
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5842794956996863e-06,
+      "loss": 0.8984,
+      "step": 8034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5821893871798376e-06,
+      "loss": 0.8309,
+      "step": 8035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.580099998908049e-06,
+      "loss": 0.8816,
+      "step": 8036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.578011331087189e-06,
+      "loss": 0.854,
+      "step": 8037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.575923383920069e-06,
+      "loss": 0.8012,
+      "step": 8038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5738361576094186e-06,
+      "loss": 0.8331,
+      "step": 8039
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5717496523578998e-06,
+      "loss": 0.8788,
+      "step": 8040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5696638683681107e-06,
+      "loss": 0.9448,
+      "step": 8041
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5675788058425723e-06,
+      "loss": 0.8338,
+      "step": 8042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.565494464983733e-06,
+      "loss": 0.8665,
+      "step": 8043
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.563410845993988e-06,
+      "loss": 0.8463,
+      "step": 8044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5613279490756426e-06,
+      "loss": 0.9107,
+      "step": 8045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5592457744309405e-06,
+      "loss": 0.8799,
+      "step": 8046
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.557164322262059e-06,
+      "loss": 0.858,
+      "step": 8047
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5550835927710982e-06,
+      "loss": 0.8313,
+      "step": 8048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.553003586160089e-06,
+      "loss": 0.8427,
+      "step": 8049
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5509243026309983e-06,
+      "loss": 0.8089,
+      "step": 8050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.548845742385717e-06,
+      "loss": 0.8814,
+      "step": 8051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.546767905626063e-06,
+      "loss": 0.8351,
+      "step": 8052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5446907925537957e-06,
+      "loss": 0.8899,
+      "step": 8053
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5426144033705937e-06,
+      "loss": 0.9254,
+      "step": 8054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5405387382780633e-06,
+      "loss": 0.8685,
+      "step": 8055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5384637974777513e-06,
+      "loss": 0.7325,
+      "step": 8056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5363895811711305e-06,
+      "loss": 0.9165,
+      "step": 8057
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5343160895595977e-06,
+      "loss": 0.8735,
+      "step": 8058
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.53224332284448e-06,
+      "loss": 0.9534,
+      "step": 8059
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.530171281227044e-06,
+      "loss": 0.7985,
+      "step": 8060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.528099964908475e-06,
+      "loss": 0.9211,
+      "step": 8061
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.52602937408989e-06,
+      "loss": 0.9295,
+      "step": 8062
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.523959508972342e-06,
+      "loss": 0.8136,
+      "step": 8063
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5218903697568075e-06,
+      "loss": 0.8477,
+      "step": 8064
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5198219566441893e-06,
+      "loss": 0.8456,
+      "step": 8065
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.517754269835332e-06,
+      "loss": 0.8422,
+      "step": 8066
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.515687309530995e-06,
+      "loss": 0.9031,
+      "step": 8067
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5136210759318814e-06,
+      "loss": 0.887,
+      "step": 8068
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5115555692386086e-06,
+      "loss": 0.8322,
+      "step": 8069
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5094907896517383e-06,
+      "loss": 0.8602,
+      "step": 8070
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.507426737371752e-06,
+      "loss": 0.771,
+      "step": 8071
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.505363412599059e-06,
+      "loss": 0.823,
+      "step": 8072
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.50330081553401e-06,
+      "loss": 0.7927,
+      "step": 8073
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5012389463768737e-06,
+      "loss": 0.7861,
+      "step": 8074
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4991778053278483e-06,
+      "loss": 0.861,
+      "step": 8075
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4971173925870694e-06,
+      "loss": 0.8765,
+      "step": 8076
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4950577083545967e-06,
+      "loss": 0.87,
+      "step": 8077
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4929987528304144e-06,
+      "loss": 0.8873,
+      "step": 8078
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.490940526214446e-06,
+      "loss": 0.8262,
+      "step": 8079
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4888830287065414e-06,
+      "loss": 0.8816,
+      "step": 8080
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.486826260506474e-06,
+      "loss": 0.7652,
+      "step": 8081
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4847702218139493e-06,
+      "loss": 0.8705,
+      "step": 8082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4827149128286064e-06,
+      "loss": 0.8895,
+      "step": 8083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.480660333750007e-06,
+      "loss": 0.8866,
+      "step": 8084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4786064847776437e-06,
+      "loss": 0.8783,
+      "step": 8085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.476553366110944e-06,
+      "loss": 0.861,
+      "step": 8086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.474500977949257e-06,
+      "loss": 0.8011,
+      "step": 8087
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4724493204918598e-06,
+      "loss": 0.8741,
+      "step": 8088
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4703983939379705e-06,
+      "loss": 0.8247,
+      "step": 8089
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4683481984867207e-06,
+      "loss": 0.8659,
+      "step": 8090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4662987343371846e-06,
+      "loss": 0.823,
+      "step": 8091
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4642500016883532e-06,
+      "loss": 0.7765,
+      "step": 8092
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.462202000739159e-06,
+      "loss": 0.9393,
+      "step": 8093
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4601547316884544e-06,
+      "loss": 0.8839,
+      "step": 8094
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4581081947350193e-06,
+      "loss": 0.8857,
+      "step": 8095
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4560623900775728e-06,
+      "loss": 0.946,
+      "step": 8096
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.454017317914753e-06,
+      "loss": 0.8541,
+      "step": 8097
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.45197297844513e-06,
+      "loss": 0.9336,
+      "step": 8098
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4499293718672067e-06,
+      "loss": 0.835,
+      "step": 8099
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.447886498379409e-06,
+      "loss": 0.8819,
+      "step": 8100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4458443581800916e-06,
+      "loss": 0.85,
+      "step": 8101
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4438029514675444e-06,
+      "loss": 0.8388,
+      "step": 8102
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4417622784399818e-06,
+      "loss": 0.8743,
+      "step": 8103
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.439722339295545e-06,
+      "loss": 0.764,
+      "step": 8104
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.43768313423231e-06,
+      "loss": 0.7976,
+      "step": 8105
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4356446634482756e-06,
+      "loss": 0.8501,
+      "step": 8106
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.433606927141368e-06,
+      "loss": 0.9033,
+      "step": 8107
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4315699255094516e-06,
+      "loss": 0.8089,
+      "step": 8108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4295336587503104e-06,
+      "loss": 0.8003,
+      "step": 8109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.427498127061658e-06,
+      "loss": 0.8219,
+      "step": 8110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4254633306411435e-06,
+      "loss": 0.8558,
+      "step": 8111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.423429269686336e-06,
+      "loss": 0.8542,
+      "step": 8112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4213959443947368e-06,
+      "loss": 0.8423,
+      "step": 8113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4193633549637765e-06,
+      "loss": 0.9247,
+      "step": 8114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4173315015908173e-06,
+      "loss": 0.8833,
+      "step": 8115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4153003844731425e-06,
+      "loss": 0.8167,
+      "step": 8116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.413270003807967e-06,
+      "loss": 0.8633,
+      "step": 8117
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.411240359792438e-06,
+      "loss": 0.8802,
+      "step": 8118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4092114526236286e-06,
+      "loss": 0.6939,
+      "step": 8119
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.407183282498534e-06,
+      "loss": 0.8663,
+      "step": 8120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4051558496140905e-06,
+      "loss": 0.7896,
+      "step": 8121
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.403129154167153e-06,
+      "loss": 0.8907,
+      "step": 8122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4011031963545052e-06,
+      "loss": 0.8547,
+      "step": 8123
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3990779763728666e-06,
+      "loss": 0.8832,
+      "step": 8124
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.397053494418875e-06,
+      "loss": 0.9285,
+      "step": 8125
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3950297506891084e-06,
+      "loss": 0.8299,
+      "step": 8126
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.393006745380061e-06,
+      "loss": 0.85,
+      "step": 8127
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.390984478688164e-06,
+      "loss": 0.8554,
+      "step": 8128
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3889629508097732e-06,
+      "loss": 0.7794,
+      "step": 8129
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.386942161941169e-06,
+      "loss": 0.8664,
+      "step": 8130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3849221122785705e-06,
+      "loss": 0.8353,
+      "step": 8131
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3829028020181154e-06,
+      "loss": 0.8224,
+      "step": 8132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.380884231355871e-06,
+      "loss": 0.8802,
+      "step": 8133
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3788664004878405e-06,
+      "loss": 0.8634,
+      "step": 8134
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3768493096099445e-06,
+      "loss": 0.8594,
+      "step": 8135
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.374832958918035e-06,
+      "loss": 0.8569,
+      "step": 8136
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.372817348607899e-06,
+      "loss": 0.8519,
+      "step": 8137
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3708024788752448e-06,
+      "loss": 0.8736,
+      "step": 8138
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3687883499157117e-06,
+      "loss": 0.826,
+      "step": 8139
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3667749619248614e-06,
+      "loss": 0.8428,
+      "step": 8140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3647623150981923e-06,
+      "loss": 0.9452,
+      "step": 8141
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3627504096311273e-06,
+      "loss": 0.8727,
+      "step": 8142
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3607392457190103e-06,
+      "loss": 0.8943,
+      "step": 8143
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3587288235571258e-06,
+      "loss": 0.8734,
+      "step": 8144
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.35671914334068e-06,
+      "loss": 0.8303,
+      "step": 8145
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.354710205264801e-06,
+      "loss": 0.9203,
+      "step": 8146
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3527020095245577e-06,
+      "loss": 0.843,
+      "step": 8147
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.350694556314934e-06,
+      "loss": 0.8572,
+      "step": 8148
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.348687845830853e-06,
+      "loss": 0.8658,
+      "step": 8149
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3466818782671597e-06,
+      "loss": 0.8668,
+      "step": 8150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3446766538186274e-06,
+      "loss": 0.7739,
+      "step": 8151
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3426721726799573e-06,
+      "loss": 0.8899,
+      "step": 8152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.340668435045775e-06,
+      "loss": 0.7634,
+      "step": 8153
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3386654411106446e-06,
+      "loss": 0.904,
+      "step": 8154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3366631910690476e-06,
+      "loss": 0.8522,
+      "step": 8155
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3346616851153935e-06,
+      "loss": 0.8275,
+      "step": 8156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3326609234440277e-06,
+      "loss": 0.858,
+      "step": 8157
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.330660906249218e-06,
+      "loss": 0.8863,
+      "step": 8158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3286616337251554e-06,
+      "loss": 0.9023,
+      "step": 8159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3266631060659685e-06,
+      "loss": 0.8209,
+      "step": 8160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.324665323465708e-06,
+      "loss": 0.834,
+      "step": 8161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.32266828611835e-06,
+      "loss": 0.8513,
+      "step": 8162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.320671994217806e-06,
+      "loss": 0.8356,
+      "step": 8163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.318676447957907e-06,
+      "loss": 0.7649,
+      "step": 8164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.316681647532413e-06,
+      "loss": 0.8948,
+      "step": 8165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3146875931350165e-06,
+      "loss": 0.8233,
+      "step": 8166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3126942849593346e-06,
+      "loss": 0.8493,
+      "step": 8167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.310701723198908e-06,
+      "loss": 0.8367,
+      "step": 8168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.308709908047213e-06,
+      "loss": 0.9191,
+      "step": 8169
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3067188396976482e-06,
+      "loss": 0.8077,
+      "step": 8170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3047285183435375e-06,
+      "loss": 0.8461,
+      "step": 8171
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3027389441781368e-06,
+      "loss": 0.8373,
+      "step": 8172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3007501173946312e-06,
+      "loss": 0.8231,
+      "step": 8173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2987620381861288e-06,
+      "loss": 0.8576,
+      "step": 8174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2967747067456614e-06,
+      "loss": 0.9266,
+      "step": 8175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2947881232662007e-06,
+      "loss": 0.8939,
+      "step": 8176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.292802287940634e-06,
+      "loss": 0.8608,
+      "step": 8177
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.290817200961779e-06,
+      "loss": 0.8671,
+      "step": 8178
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2888328625223855e-06,
+      "loss": 0.887,
+      "step": 8179
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.286849272815126e-06,
+      "loss": 0.8975,
+      "step": 8180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2848664320325973e-06,
+      "loss": 0.8942,
+      "step": 8181
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2828843403673338e-06,
+      "loss": 0.8163,
+      "step": 8182
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2809029980117857e-06,
+      "loss": 0.8358,
+      "step": 8183
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2789224051583403e-06,
+      "loss": 0.8475,
+      "step": 8184
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2769425619993022e-06,
+      "loss": 0.8383,
+      "step": 8185
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.274963468726914e-06,
+      "loss": 0.8727,
+      "step": 8186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.272985125533338e-06,
+      "loss": 0.7763,
+      "step": 8187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2710075326106618e-06,
+      "loss": 0.8684,
+      "step": 8188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2690306901509095e-06,
+      "loss": 0.8326,
+      "step": 8189
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2670545983460245e-06,
+      "loss": 0.8332,
+      "step": 8190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2650792573878765e-06,
+      "loss": 0.7934,
+      "step": 8191
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.263104667468272e-06,
+      "loss": 0.844,
+      "step": 8192
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2611308287789347e-06,
+      "loss": 0.8446,
+      "step": 8193
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.259157741511515e-06,
+      "loss": 0.7811,
+      "step": 8194
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.257185405857597e-06,
+      "loss": 0.8529,
+      "step": 8195
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2552138220086927e-06,
+      "loss": 0.9175,
+      "step": 8196
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2532429901562337e-06,
+      "loss": 0.8167,
+      "step": 8197
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2512729104915787e-06,
+      "loss": 0.8846,
+      "step": 8198
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.249303583206024e-06,
+      "loss": 0.8994,
+      "step": 8199
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2473350084907806e-06,
+      "loss": 0.8756,
+      "step": 8200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.245367186536991e-06,
+      "loss": 0.9422,
+      "step": 8201
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.243400117535729e-06,
+      "loss": 0.7682,
+      "step": 8202
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.24143380167799e-06,
+      "loss": 0.7991,
+      "step": 8203
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2394682391546928e-06,
+      "loss": 0.8538,
+      "step": 8204
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2375034301566944e-06,
+      "loss": 0.8404,
+      "step": 8205
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2355393748747702e-06,
+      "loss": 0.8656,
+      "step": 8206
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.233576073499617e-06,
+      "loss": 0.8439,
+      "step": 8207
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2316135262218787e-06,
+      "loss": 0.8645,
+      "step": 8208
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2296517332321067e-06,
+      "loss": 0.9009,
+      "step": 8209
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.227690694720784e-06,
+      "loss": 0.8599,
+      "step": 8210
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2257304108783216e-06,
+      "loss": 0.8428,
+      "step": 8211
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.223770881895061e-06,
+      "loss": 0.808,
+      "step": 8212
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2218121079612655e-06,
+      "loss": 0.9089,
+      "step": 8213
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2198540892671215e-06,
+      "loss": 0.9106,
+      "step": 8214
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2178968260027547e-06,
+      "loss": 0.8276,
+      "step": 8215
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.215940318358206e-06,
+      "loss": 0.7903,
+      "step": 8216
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.213984566523444e-06,
+      "loss": 0.8241,
+      "step": 8217
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2120295706883698e-06,
+      "loss": 0.8897,
+      "step": 8218
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.21007533104281e-06,
+      "loss": 0.8579,
+      "step": 8219
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2081218477765097e-06,
+      "loss": 0.8807,
+      "step": 8220
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.206169121079154e-06,
+      "loss": 0.8855,
+      "step": 8221
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.204217151140342e-06,
+      "loss": 0.8827,
+      "step": 8222
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2022659381496025e-06,
+      "loss": 0.7333,
+      "step": 8223
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.200315482296398e-06,
+      "loss": 0.8167,
+      "step": 8224
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.19836578377011e-06,
+      "loss": 0.8411,
+      "step": 8225
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1964168427600462e-06,
+      "loss": 0.88,
+      "step": 8226
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1944686594554478e-06,
+      "loss": 0.8906,
+      "step": 8227
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1925212340454737e-06,
+      "loss": 0.8115,
+      "step": 8228
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1905745667192137e-06,
+      "loss": 0.8304,
+      "step": 8229
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1886286576656834e-06,
+      "loss": 0.9259,
+      "step": 8230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.18668350707383e-06,
+      "loss": 0.8824,
+      "step": 8231
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.184739115132517e-06,
+      "loss": 0.8677,
+      "step": 8232
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1827954820305385e-06,
+      "loss": 0.8287,
+      "step": 8233
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1808526079566215e-06,
+      "loss": 0.8796,
+      "step": 8234
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.178910493099409e-06,
+      "loss": 0.8254,
+      "step": 8235
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1769691376474722e-06,
+      "loss": 0.8149,
+      "step": 8236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.175028541789317e-06,
+      "loss": 0.8198,
+      "step": 8237
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1730887057133678e-06,
+      "loss": 0.889,
+      "step": 8238
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1711496296079725e-06,
+      "loss": 0.9229,
+      "step": 8239
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1692113136614177e-06,
+      "loss": 0.8846,
+      "step": 8240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.167273758061904e-06,
+      "loss": 0.8881,
+      "step": 8241
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1653369629975595e-06,
+      "loss": 0.8658,
+      "step": 8242
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.163400928656445e-06,
+      "loss": 0.8926,
+      "step": 8243
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1614656552265457e-06,
+      "loss": 0.8508,
+      "step": 8244
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.15953114289577e-06,
+      "loss": 0.9288,
+      "step": 8245
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.157597391851949e-06,
+      "loss": 0.8411,
+      "step": 8246
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.15566440228285e-06,
+      "loss": 0.8639,
+      "step": 8247
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1537321743761587e-06,
+      "loss": 0.8063,
+      "step": 8248
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.151800708319486e-06,
+      "loss": 0.8687,
+      "step": 8249
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1498700043003773e-06,
+      "loss": 0.7911,
+      "step": 8250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.147940062506295e-06,
+      "loss": 0.9131,
+      "step": 8251
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1460108831246295e-06,
+      "loss": 0.8225,
+      "step": 8252
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1440824663426994e-06,
+      "loss": 0.7925,
+      "step": 8253
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.142154812347753e-06,
+      "loss": 0.8807,
+      "step": 8254
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1402279213269562e-06,
+      "loss": 0.8596,
+      "step": 8255
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1383017934674012e-06,
+      "loss": 0.8274,
+      "step": 8256
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.136376428956116e-06,
+      "loss": 0.8748,
+      "step": 8257
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1344518279800454e-06,
+      "loss": 0.9251,
+      "step": 8258
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.132527990726059e-06,
+      "loss": 0.7452,
+      "step": 8259
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.130604917380962e-06,
+      "loss": 0.8994,
+      "step": 8260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.128682608131476e-06,
+      "loss": 0.7767,
+      "step": 8261
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.12676106316425e-06,
+      "loss": 0.9135,
+      "step": 8262
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1248402826658642e-06,
+      "loss": 0.8668,
+      "step": 8263
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1229202668228197e-06,
+      "loss": 0.8723,
+      "step": 8264
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.121001015821539e-06,
+      "loss": 0.8799,
+      "step": 8265
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1190825298483855e-06,
+      "loss": 0.8467,
+      "step": 8266
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1171648090896347e-06,
+      "loss": 0.8773,
+      "step": 8267
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.115247853731488e-06,
+      "loss": 0.8303,
+      "step": 8268
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.113331663960081e-06,
+      "loss": 0.8253,
+      "step": 8269
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.11141623996147e-06,
+      "loss": 0.8652,
+      "step": 8270
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1095015819216345e-06,
+      "loss": 0.8339,
+      "step": 8271
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.107587690026481e-06,
+      "loss": 0.905,
+      "step": 8272
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1056745644618494e-06,
+      "loss": 0.8562,
+      "step": 8273
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.103762205413493e-06,
+      "loss": 0.8447,
+      "step": 8274
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1018506130670956e-06,
+      "loss": 0.8594,
+      "step": 8275
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0999397876082726e-06,
+      "loss": 0.7711,
+      "step": 8276
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0980297292225548e-06,
+      "loss": 0.8156,
+      "step": 8277
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.096120438095404e-06,
+      "loss": 0.8443,
+      "step": 8278
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.094211914412212e-06,
+      "loss": 0.8039,
+      "step": 8279
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.092304158358286e-06,
+      "loss": 0.8256,
+      "step": 8280
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.090397170118863e-06,
+      "loss": 0.7912,
+      "step": 8281
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0884909498791106e-06,
+      "loss": 0.8369,
+      "step": 8282
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0865854978241142e-06,
+      "loss": 0.8903,
+      "step": 8283
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0846808141388852e-06,
+      "loss": 0.8487,
+      "step": 8284
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0827768990083686e-06,
+      "loss": 0.789,
+      "step": 8285
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.080873752617426e-06,
+      "loss": 0.9234,
+      "step": 8286
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0789713751508455e-06,
+      "loss": 0.9608,
+      "step": 8287
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0770697667933436e-06,
+      "loss": 0.9746,
+      "step": 8288
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.075168927729566e-06,
+      "loss": 0.7896,
+      "step": 8289
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.073268858144074e-06,
+      "loss": 0.8461,
+      "step": 8290
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0713695582213578e-06,
+      "loss": 0.9257,
+      "step": 8291
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0694710281458372e-06,
+      "loss": 0.841,
+      "step": 8292
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0675732681018535e-06,
+      "loss": 0.844,
+      "step": 8293
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0656762782736693e-06,
+      "loss": 0.9359,
+      "step": 8294
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0637800588454836e-06,
+      "loss": 0.8075,
+      "step": 8295
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.061884610001411e-06,
+      "loss": 0.9173,
+      "step": 8296
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0599899319254913e-06,
+      "loss": 0.8519,
+      "step": 8297
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0580960248016966e-06,
+      "loss": 0.8922,
+      "step": 8298
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0562028888139184e-06,
+      "loss": 0.8535,
+      "step": 8299
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0543105241459713e-06,
+      "loss": 0.8369,
+      "step": 8300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0524189309816014e-06,
+      "loss": 0.8119,
+      "step": 8301
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0505281095044804e-06,
+      "loss": 0.8281,
+      "step": 8302
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0486380598981982e-06,
+      "loss": 0.868,
+      "step": 8303
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0467487823462696e-06,
+      "loss": 0.8096,
+      "step": 8304
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.044860277032145e-06,
+      "loss": 0.88,
+      "step": 8305
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.042972544139189e-06,
+      "loss": 0.7791,
+      "step": 8306
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.041085583850694e-06,
+      "loss": 0.815,
+      "step": 8307
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.039199396349881e-06,
+      "loss": 0.9137,
+      "step": 8308
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0373139818198938e-06,
+      "loss": 0.8411,
+      "step": 8309
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0354293404437963e-06,
+      "loss": 0.8877,
+      "step": 8310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.033545472404587e-06,
+      "loss": 0.8746,
+      "step": 8311
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0316623778851784e-06,
+      "loss": 0.8273,
+      "step": 8312
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.02978005706842e-06,
+      "loss": 0.856,
+      "step": 8313
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0278985101370753e-06,
+      "loss": 0.8915,
+      "step": 8314
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.02601773727384e-06,
+      "loss": 0.9161,
+      "step": 8315
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.024137738661329e-06,
+      "loss": 0.8362,
+      "step": 8316
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.022258514482085e-06,
+      "loss": 0.8706,
+      "step": 8317
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.020380064918579e-06,
+      "loss": 0.8548,
+      "step": 8318
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0185023901531984e-06,
+      "loss": 0.8416,
+      "step": 8319
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.01662549036826e-06,
+      "loss": 0.876,
+      "step": 8320
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0147493657460105e-06,
+      "loss": 0.815,
+      "step": 8321
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0128740164686134e-06,
+      "loss": 0.8351,
+      "step": 8322
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0109994427181524e-06,
+      "loss": 0.8221,
+      "step": 8323
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.009125644676656e-06,
+      "loss": 0.8398,
+      "step": 8324
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0072526225260592e-06,
+      "loss": 0.8551,
+      "step": 8325
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0053803764482226e-06,
+      "loss": 0.8607,
+      "step": 8326
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0035089066249437e-06,
+      "loss": 0.8517,
+      "step": 8327
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.001638213237932e-06,
+      "loss": 0.8737,
+      "step": 8328
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9997682964688248e-06,
+      "loss": 0.8232,
+      "step": 8329
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.997899156499191e-06,
+      "loss": 0.8731,
+      "step": 8330
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.996030793510516e-06,
+      "loss": 0.8632,
+      "step": 8331
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.994163207684212e-06,
+      "loss": 0.8624,
+      "step": 8332
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.992296399201615e-06,
+      "loss": 0.8658,
+      "step": 8333
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9904303682439896e-06,
+      "loss": 0.8774,
+      "step": 8334
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.988565114992519e-06,
+      "loss": 0.8467,
+      "step": 8335
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986700639628316e-06,
+      "loss": 0.822,
+      "step": 8336
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9848369423324177e-06,
+      "loss": 0.8481,
+      "step": 8337
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9829740232857807e-06,
+      "loss": 0.825,
+      "step": 8338
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9811118826692887e-06,
+      "loss": 0.7901,
+      "step": 8339
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9792505206637523e-06,
+      "loss": 0.7969,
+      "step": 8340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.977389937449905e-06,
+      "loss": 0.8989,
+      "step": 8341
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9755301332083997e-06,
+      "loss": 0.883,
+      "step": 8342
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9736711081198235e-06,
+      "loss": 0.8442,
+      "step": 8343
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9718128623646792e-06,
+      "loss": 0.7952,
+      "step": 8344
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.969955396123395e-06,
+      "loss": 0.8263,
+      "step": 8345
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9680987095763315e-06,
+      "loss": 0.8769,
+      "step": 8346
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9662428029037617e-06,
+      "loss": 0.8064,
+      "step": 8347
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.964387676285894e-06,
+      "loss": 0.8261,
+      "step": 8348
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9625333299028494e-06,
+      "loss": 0.8693,
+      "step": 8349
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9606797639346874e-06,
+      "loss": 0.8969,
+      "step": 8350
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.95882697856138e-06,
+      "loss": 0.9421,
+      "step": 8351
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9569749739628243e-06,
+      "loss": 0.8797,
+      "step": 8352
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9551237503188504e-06,
+      "loss": 0.843,
+      "step": 8353
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9532733078092034e-06,
+      "loss": 0.8716,
+      "step": 8354
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.951423646613555e-06,
+      "loss": 0.8412,
+      "step": 8355
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9495747669115062e-06,
+      "loss": 0.8856,
+      "step": 8356
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9477266688825746e-06,
+      "loss": 0.8612,
+      "step": 8357
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9458793527062035e-06,
+      "loss": 0.8289,
+      "step": 8358
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.944032818561764e-06,
+      "loss": 0.8196,
+      "step": 8359
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9421870666285523e-06,
+      "loss": 0.8314,
+      "step": 8360
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9403420970857833e-06,
+      "loss": 0.8588,
+      "step": 8361
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9384979101125944e-06,
+      "loss": 0.8903,
+      "step": 8362
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.936654505888058e-06,
+      "loss": 0.8433,
+      "step": 8363
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.934811884591159e-06,
+      "loss": 0.9137,
+      "step": 8364
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9329700464008083e-06,
+      "loss": 0.8336,
+      "step": 8365
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.93112899149585e-06,
+      "loss": 0.856,
+      "step": 8366
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9292887200550394e-06,
+      "loss": 0.8598,
+      "step": 8367
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9274492322570616e-06,
+      "loss": 0.8531,
+      "step": 8368
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9256105282805304e-06,
+      "loss": 0.8425,
+      "step": 8369
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.923772608303972e-06,
+      "loss": 0.8266,
+      "step": 8370
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9219354725058503e-06,
+      "loss": 0.8459,
+      "step": 8371
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9200991210645394e-06,
+      "loss": 0.8568,
+      "step": 8372
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9182635541583495e-06,
+      "loss": 0.8701,
+      "step": 8373
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.916428771965506e-06,
+      "loss": 0.913,
+      "step": 8374
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.914594774664159e-06,
+      "loss": 0.8875,
+      "step": 8375
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.912761562432388e-06,
+      "loss": 0.8111,
+      "step": 8376
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.910929135448192e-06,
+      "loss": 0.9042,
+      "step": 8377
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9090974938894902e-06,
+      "loss": 0.8727,
+      "step": 8378
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9072666379341354e-06,
+      "loss": 0.8264,
+      "step": 8379
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9054365677598963e-06,
+      "loss": 0.8448,
+      "step": 8380
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.903607283544464e-06,
+      "loss": 0.9162,
+      "step": 8381
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9017787854654613e-06,
+      "loss": 0.817,
+      "step": 8382
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.89995107370043e-06,
+      "loss": 0.8428,
+      "step": 8383
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.898124148426832e-06,
+      "loss": 0.85,
+      "step": 8384
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8962980098220618e-06,
+      "loss": 0.904,
+      "step": 8385
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8944726580634287e-06,
+      "loss": 0.8549,
+      "step": 8386
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8926480933281676e-06,
+      "loss": 0.8837,
+      "step": 8387
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8908243157934424e-06,
+      "loss": 0.8782,
+      "step": 8388
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8890013256363349e-06,
+      "loss": 0.8589,
+      "step": 8389
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8871791230338499e-06,
+      "loss": 0.8407,
+      "step": 8390
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8853577081629226e-06,
+      "loss": 0.8174,
+      "step": 8391
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.883537081200404e-06,
+      "loss": 0.9035,
+      "step": 8392
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8817172423230713e-06,
+      "loss": 0.8465,
+      "step": 8393
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8798981917076254e-06,
+      "loss": 0.8808,
+      "step": 8394
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8780799295306952e-06,
+      "loss": 0.8558,
+      "step": 8395
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.876262455968826e-06,
+      "loss": 0.8733,
+      "step": 8396
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8744457711984864e-06,
+      "loss": 0.8463,
+      "step": 8397
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.872629875396076e-06,
+      "loss": 0.8301,
+      "step": 8398
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.870814768737912e-06,
+      "loss": 0.8578,
+      "step": 8399
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8690004514002314e-06,
+      "loss": 0.9194,
+      "step": 8400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8671869235592054e-06,
+      "loss": 0.8143,
+      "step": 8401
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8653741853909201e-06,
+      "loss": 0.8268,
+      "step": 8402
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8635622370713847e-06,
+      "loss": 0.819,
+      "step": 8403
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.861751078776538e-06,
+      "loss": 0.8863,
+      "step": 8404
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8599407106822332e-06,
+      "loss": 0.8928,
+      "step": 8405
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8581311329642592e-06,
+      "loss": 0.8105,
+      "step": 8406
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8563223457983137e-06,
+      "loss": 0.8917,
+      "step": 8407
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8545143493600293e-06,
+      "loss": 0.8526,
+      "step": 8408
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8527071438249566e-06,
+      "loss": 0.8786,
+      "step": 8409
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8509007293685666e-06,
+      "loss": 0.7454,
+      "step": 8410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.849095106166262e-06,
+      "loss": 0.7981,
+      "step": 8411
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8472902743933608e-06,
+      "loss": 0.834,
+      "step": 8412
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8454862342251045e-06,
+      "loss": 0.8795,
+      "step": 8413
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8436829858366655e-06,
+      "loss": 0.849,
+      "step": 8414
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8418805294031317e-06,
+      "loss": 0.9039,
+      "step": 8415
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8400788650995137e-06,
+      "loss": 0.9318,
+      "step": 8416
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8382779931007511e-06,
+      "loss": 0.8926,
+      "step": 8417
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8364779135817045e-06,
+      "loss": 0.8929,
+      "step": 8418
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8346786267171546e-06,
+      "loss": 0.8765,
+      "step": 8419
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8328801326818045e-06,
+      "loss": 0.8428,
+      "step": 8420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8310824316502884e-06,
+      "loss": 0.8293,
+      "step": 8421
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.829285523797155e-06,
+      "loss": 0.9128,
+      "step": 8422
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8274894092968754e-06,
+      "loss": 0.8327,
+      "step": 8423
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8256940883238538e-06,
+      "loss": 0.8321,
+      "step": 8424
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8238995610524068e-06,
+      "loss": 0.8134,
+      "step": 8425
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.822105827656776e-06,
+      "loss": 0.8751,
+      "step": 8426
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8203128883111331e-06,
+      "loss": 0.8977,
+      "step": 8427
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8185207431895613e-06,
+      "loss": 0.8793,
+      "step": 8428
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8167293924660778e-06,
+      "loss": 0.9495,
+      "step": 8429
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8149388363146148e-06,
+      "loss": 0.9122,
+      "step": 8430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.813149074909032e-06,
+      "loss": 0.8022,
+      "step": 8431
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8113601084231091e-06,
+      "loss": 0.7966,
+      "step": 8432
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8095719370305476e-06,
+      "loss": 0.8726,
+      "step": 8433
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8077845609049782e-06,
+      "loss": 0.8801,
+      "step": 8434
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8059979802199478e-06,
+      "loss": 0.9184,
+      "step": 8435
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8042121951489254e-06,
+      "loss": 0.7968,
+      "step": 8436
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8024272058653102e-06,
+      "loss": 0.7916,
+      "step": 8437
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.800643012542418e-06,
+      "loss": 0.8727,
+      "step": 8438
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7988596153534855e-06,
+      "loss": 0.8461,
+      "step": 8439
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7970770144716777e-06,
+      "loss": 0.8618,
+      "step": 8440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7952952100700837e-06,
+      "loss": 0.8353,
+      "step": 8441
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7935142023217056e-06,
+      "loss": 0.866,
+      "step": 8442
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7917339913994803e-06,
+      "loss": 0.7708,
+      "step": 8443
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7899545774762573e-06,
+      "loss": 0.8549,
+      "step": 8444
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7881759607248094e-06,
+      "loss": 0.8846,
+      "step": 8445
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7863981413178433e-06,
+      "loss": 0.8646,
+      "step": 8446
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.784621119427975e-06,
+      "loss": 0.8962,
+      "step": 8447
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7828448952277456e-06,
+      "loss": 0.8601,
+      "step": 8448
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7810694688896279e-06,
+      "loss": 0.8782,
+      "step": 8449
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7792948405860079e-06,
+      "loss": 0.8202,
+      "step": 8450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7775210104891938e-06,
+      "loss": 0.9075,
+      "step": 8451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7757479787714217e-06,
+      "loss": 0.8651,
+      "step": 8452
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.77397574560485e-06,
+      "loss": 0.7769,
+      "step": 8453
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7722043111615572e-06,
+      "loss": 0.8668,
+      "step": 8454
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.77043367561354e-06,
+      "loss": 0.8399,
+      "step": 8455
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.768663839132727e-06,
+      "loss": 0.8613,
+      "step": 8456
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7668948018909627e-06,
+      "loss": 0.8708,
+      "step": 8457
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7651265640600113e-06,
+      "loss": 0.908,
+      "step": 8458
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7633591258115713e-06,
+      "loss": 0.8446,
+      "step": 8459
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7615924873172506e-06,
+      "loss": 0.8128,
+      "step": 8460
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7598266487485848e-06,
+      "loss": 0.7951,
+      "step": 8461
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7580616102770353e-06,
+      "loss": 0.7382,
+      "step": 8462
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7562973720739763e-06,
+      "loss": 0.8712,
+      "step": 8463
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.754533934310717e-06,
+      "loss": 0.7961,
+      "step": 8464
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7527712971584777e-06,
+      "loss": 0.8436,
+      "step": 8465
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7510094607884075e-06,
+      "loss": 0.7901,
+      "step": 8466
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7492484253715757e-06,
+      "loss": 0.9498,
+      "step": 8467
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7474881910789698e-06,
+      "loss": 0.7558,
+      "step": 8468
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7457287580815096e-06,
+      "loss": 0.8517,
+      "step": 8469
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7439701265500274e-06,
+      "loss": 0.8357,
+      "step": 8470
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.74221229665528e-06,
+      "loss": 0.8608,
+      "step": 8471
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.740455268567951e-06,
+      "loss": 0.8649,
+      "step": 8472
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7386990424586426e-06,
+      "loss": 0.8585,
+      "step": 8473
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7369436184978738e-06,
+      "loss": 0.8798,
+      "step": 8474
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7351889968560964e-06,
+      "loss": 0.8934,
+      "step": 8475
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7334351777036807e-06,
+      "loss": 0.8513,
+      "step": 8476
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7316821612109136e-06,
+      "loss": 0.8534,
+      "step": 8477
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.729929947548008e-06,
+      "loss": 0.8666,
+      "step": 8478
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7281785368851023e-06,
+      "loss": 0.874,
+      "step": 8479
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7264279293922503e-06,
+      "loss": 0.9206,
+      "step": 8480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7246781252394295e-06,
+      "loss": 0.8373,
+      "step": 8481
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7229291245965463e-06,
+      "loss": 0.9149,
+      "step": 8482
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7211809276334202e-06,
+      "loss": 0.7896,
+      "step": 8483
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7194335345197933e-06,
+      "loss": 0.8204,
+      "step": 8484
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7176869454253387e-06,
+      "loss": 0.856,
+      "step": 8485
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7159411605196407e-06,
+      "loss": 0.9069,
+      "step": 8486
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7141961799722107e-06,
+      "loss": 0.7518,
+      "step": 8487
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7124520039524805e-06,
+      "loss": 0.9224,
+      "step": 8488
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7107086326298083e-06,
+      "loss": 0.8489,
+      "step": 8489
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7089660661734685e-06,
+      "loss": 0.8278,
+      "step": 8490
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.707224304752656e-06,
+      "loss": 0.8036,
+      "step": 8491
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7054833485364962e-06,
+      "loss": 0.8681,
+      "step": 8492
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.703743197694029e-06,
+      "loss": 0.8434,
+      "step": 8493
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.702003852394214e-06,
+      "loss": 0.8077,
+      "step": 8494
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7002653128059432e-06,
+      "loss": 0.7782,
+      "step": 8495
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6985275790980205e-06,
+      "loss": 0.8033,
+      "step": 8496
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6967906514391729e-06,
+      "loss": 0.8256,
+      "step": 8497
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6950545299980526e-06,
+      "loss": 0.8819,
+      "step": 8498
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6933192149432343e-06,
+      "loss": 0.8631,
+      "step": 8499
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.691584706443209e-06,
+      "loss": 0.9073,
+      "step": 8500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6898510046663962e-06,
+      "loss": 0.8575,
+      "step": 8501
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6881181097811305e-06,
+      "loss": 0.8325,
+      "step": 8502
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6863860219556694e-06,
+      "loss": 0.929,
+      "step": 8503
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6846547413581981e-06,
+      "loss": 0.8701,
+      "step": 8504
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6829242681568158e-06,
+      "loss": 0.8319,
+      "step": 8505
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6811946025195459e-06,
+      "loss": 0.8567,
+      "step": 8506
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6794657446143382e-06,
+      "loss": 0.8804,
+      "step": 8507
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6777376946090552e-06,
+      "loss": 0.8419,
+      "step": 8508
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.676010452671486e-06,
+      "loss": 0.8162,
+      "step": 8509
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.674284018969342e-06,
+      "loss": 0.8669,
+      "step": 8510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6725583936702572e-06,
+      "loss": 0.8221,
+      "step": 8511
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6708335769417827e-06,
+      "loss": 0.8296,
+      "step": 8512
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6691095689513914e-06,
+      "loss": 0.8831,
+      "step": 8513
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.667386369866484e-06,
+      "loss": 0.8665,
+      "step": 8514
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.665663979854375e-06,
+      "loss": 0.8253,
+      "step": 8515
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6639423990823011e-06,
+      "loss": 0.8429,
+      "step": 8516
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6622216277174296e-06,
+      "loss": 0.7938,
+      "step": 8517
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.660501665926838e-06,
+      "loss": 0.8401,
+      "step": 8518
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.658782513877527e-06,
+      "loss": 0.9003,
+      "step": 8519
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6570641717364277e-06,
+      "loss": 0.8593,
+      "step": 8520
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6553466396703833e-06,
+      "loss": 0.7902,
+      "step": 8521
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.653629917846159e-06,
+      "loss": 0.8635,
+      "step": 8522
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.651914006430445e-06,
+      "loss": 0.8604,
+      "step": 8523
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6501989055898537e-06,
+      "loss": 0.8197,
+      "step": 8524
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.648484615490915e-06,
+      "loss": 0.8059,
+      "step": 8525
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6467711363000794e-06,
+      "loss": 0.8788,
+      "step": 8526
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6450584681837245e-06,
+      "loss": 0.8806,
+      "step": 8527
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6433466113081442e-06,
+      "loss": 0.8816,
+      "step": 8528
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6416355658395511e-06,
+      "loss": 0.8781,
+      "step": 8529
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6399253319440888e-06,
+      "loss": 0.8264,
+      "step": 8530
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6382159097878125e-06,
+      "loss": 0.8362,
+      "step": 8531
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6365072995367004e-06,
+      "loss": 0.8561,
+      "step": 8532
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6347995013566576e-06,
+      "loss": 0.8108,
+      "step": 8533
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6330925154135057e-06,
+      "loss": 0.8401,
+      "step": 8534
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.631386341872988e-06,
+      "loss": 0.7948,
+      "step": 8535
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6296809809007652e-06,
+      "loss": 0.8903,
+      "step": 8536
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6279764326624293e-06,
+      "loss": 0.8307,
+      "step": 8537
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6262726973234844e-06,
+      "loss": 0.7794,
+      "step": 8538
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6245697750493539e-06,
+      "loss": 0.8586,
+      "step": 8539
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6228676660053932e-06,
+      "loss": 0.856,
+      "step": 8540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6211663703568702e-06,
+      "loss": 0.8146,
+      "step": 8541
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6194658882689718e-06,
+      "loss": 0.8738,
+      "step": 8542
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6177662199068167e-06,
+      "loss": 0.8183,
+      "step": 8543
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6160673654354331e-06,
+      "loss": 0.8763,
+      "step": 8544
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6143693250197722e-06,
+      "loss": 0.8015,
+      "step": 8545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6126720988247168e-06,
+      "loss": 0.8545,
+      "step": 8546
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6109756870150595e-06,
+      "loss": 0.8215,
+      "step": 8547
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.609280089755515e-06,
+      "loss": 0.8511,
+      "step": 8548
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6075853072107194e-06,
+      "loss": 0.855,
+      "step": 8549
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.605891339545237e-06,
+      "loss": 0.8945,
+      "step": 8550
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6041981869235446e-06,
+      "loss": 0.8987,
+      "step": 8551
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6025058495100388e-06,
+      "loss": 0.8172,
+      "step": 8552
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6008143274690457e-06,
+      "loss": 0.8374,
+      "step": 8553
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5991236209648052e-06,
+      "loss": 0.8961,
+      "step": 8554
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5974337301614785e-06,
+      "loss": 0.818,
+      "step": 8555
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5957446552231526e-06,
+      "loss": 0.8878,
+      "step": 8556
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5940563963138267e-06,
+      "loss": 0.7899,
+      "step": 8557
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5923689535974307e-06,
+      "loss": 0.9164,
+      "step": 8558
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5906823272378102e-06,
+      "loss": 0.8566,
+      "step": 8559
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.588996517398731e-06,
+      "loss": 0.87,
+      "step": 8560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5873115242438774e-06,
+      "loss": 0.9444,
+      "step": 8561
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5856273479368611e-06,
+      "loss": 0.8311,
+      "step": 8562
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.583943988641211e-06,
+      "loss": 0.7941,
+      "step": 8563
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.582261446520371e-06,
+      "loss": 0.771,
+      "step": 8564
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5805797217377182e-06,
+      "loss": 0.8777,
+      "step": 8565
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5788988144565397e-06,
+      "loss": 0.85,
+      "step": 8566
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.577218724840046e-06,
+      "loss": 0.8341,
+      "step": 8567
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.575539453051369e-06,
+      "loss": 0.8344,
+      "step": 8568
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5738609992535648e-06,
+      "loss": 0.8027,
+      "step": 8569
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.572183363609603e-06,
+      "loss": 0.8431,
+      "step": 8570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5705065462823765e-06,
+      "loss": 0.8553,
+      "step": 8571
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.568830547434703e-06,
+      "loss": 0.851,
+      "step": 8572
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5671553672293149e-06,
+      "loss": 0.7899,
+      "step": 8573
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5654810058288661e-06,
+      "loss": 0.9084,
+      "step": 8574
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5638074633959354e-06,
+      "loss": 0.8322,
+      "step": 8575
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5621347400930176e-06,
+      "loss": 0.8906,
+      "step": 8576
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5604628360825268e-06,
+      "loss": 0.7986,
+      "step": 8577
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5587917515268048e-06,
+      "loss": 0.8756,
+      "step": 8578
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5571214865881057e-06,
+      "loss": 0.9067,
+      "step": 8579
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5554520414286067e-06,
+      "loss": 0.9064,
+      "step": 8580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.553783416210408e-06,
+      "loss": 0.8474,
+      "step": 8581
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5521156110955293e-06,
+      "loss": 0.7972,
+      "step": 8582
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5504486262459096e-06,
+      "loss": 0.797,
+      "step": 8583
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5487824618234049e-06,
+      "loss": 0.8796,
+      "step": 8584
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5471171179897982e-06,
+      "loss": 0.9468,
+      "step": 8585
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.54545259490679e-06,
+      "loss": 0.8331,
+      "step": 8586
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5437888927359968e-06,
+      "loss": 0.7693,
+      "step": 8587
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5421260116389636e-06,
+      "loss": 0.7886,
+      "step": 8588
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5404639517771503e-06,
+      "loss": 0.8127,
+      "step": 8589
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5388027133119343e-06,
+      "loss": 0.8893,
+      "step": 8590
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5371422964046223e-06,
+      "loss": 0.8519,
+      "step": 8591
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.535482701216433e-06,
+      "loss": 0.9082,
+      "step": 8592
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5338239279085099e-06,
+      "loss": 0.9149,
+      "step": 8593
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5321659766419129e-06,
+      "loss": 0.8765,
+      "step": 8594
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5305088475776286e-06,
+      "loss": 0.7662,
+      "step": 8595
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5288525408765564e-06,
+      "loss": 0.8654,
+      "step": 8596
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5271970566995165e-06,
+      "loss": 0.8477,
+      "step": 8597
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5255423952072567e-06,
+      "loss": 0.7983,
+      "step": 8598
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5238885565604378e-06,
+      "loss": 0.782,
+      "step": 8599
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.52223554091964e-06,
+      "loss": 0.9261,
+      "step": 8600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5205833484453715e-06,
+      "loss": 0.8903,
+      "step": 8601
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5189319792980517e-06,
+      "loss": 0.8528,
+      "step": 8602
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5172814336380216e-06,
+      "loss": 0.7707,
+      "step": 8603
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5156317116255515e-06,
+      "loss": 0.8944,
+      "step": 8604
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5139828134208213e-06,
+      "loss": 0.8938,
+      "step": 8605
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5123347391839305e-06,
+      "loss": 0.9434,
+      "step": 8606
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.510687489074909e-06,
+      "loss": 0.8326,
+      "step": 8607
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5090410632536968e-06,
+      "loss": 0.8956,
+      "step": 8608
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5073954618801567e-06,
+      "loss": 0.8486,
+      "step": 8609
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5057506851140701e-06,
+      "loss": 0.8824,
+      "step": 8610
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5041067331151439e-06,
+      "loss": 0.8572,
+      "step": 8611
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5024636060429998e-06,
+      "loss": 0.8361,
+      "step": 8612
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5008213040571784e-06,
+      "loss": 0.8816,
+      "step": 8613
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4991798273171465e-06,
+      "loss": 0.8718,
+      "step": 8614
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.497539175982282e-06,
+      "loss": 0.9115,
+      "step": 8615
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.49589935021189e-06,
+      "loss": 0.9214,
+      "step": 8616
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4942603501651964e-06,
+      "loss": 0.8528,
+      "step": 8617
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4926221760013393e-06,
+      "loss": 0.8965,
+      "step": 8618
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.490984827879378e-06,
+      "loss": 0.9229,
+      "step": 8619
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4893483059583014e-06,
+      "loss": 0.8904,
+      "step": 8620
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4877126103970064e-06,
+      "loss": 0.8553,
+      "step": 8621
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4860777413543138e-06,
+      "loss": 0.84,
+      "step": 8622
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4844436989889678e-06,
+      "loss": 0.8649,
+      "step": 8623
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4828104834596268e-06,
+      "loss": 0.8525,
+      "step": 8624
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.481178094924871e-06,
+      "loss": 0.8098,
+      "step": 8625
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4795465335432036e-06,
+      "loss": 0.817,
+      "step": 8626
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4779157994730398e-06,
+      "loss": 0.8513,
+      "step": 8627
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4762858928727241e-06,
+      "loss": 0.7468,
+      "step": 8628
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4746568139005125e-06,
+      "loss": 0.825,
+      "step": 8629
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4730285627145858e-06,
+      "loss": 0.8112,
+      "step": 8630
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4714011394730431e-06,
+      "loss": 0.9006,
+      "step": 8631
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4697745443338984e-06,
+      "loss": 0.8506,
+      "step": 8632
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.468148777455094e-06,
+      "loss": 0.8736,
+      "step": 8633
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4665238389944859e-06,
+      "loss": 0.8552,
+      "step": 8634
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4648997291098465e-06,
+      "loss": 0.794,
+      "step": 8635
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.46327644795888e-06,
+      "loss": 0.8506,
+      "step": 8636
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4616539956991982e-06,
+      "loss": 0.8837,
+      "step": 8637
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4600323724883337e-06,
+      "loss": 0.8513,
+      "step": 8638
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.458411578483745e-06,
+      "loss": 0.8193,
+      "step": 8639
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4567916138428072e-06,
+      "loss": 0.8751,
+      "step": 8640
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4551724787228138e-06,
+      "loss": 0.9077,
+      "step": 8641
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4535541732809755e-06,
+      "loss": 0.842,
+      "step": 8642
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4519366976744286e-06,
+      "loss": 0.8373,
+      "step": 8643
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4503200520602245e-06,
+      "loss": 0.7172,
+      "step": 8644
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4487042365953309e-06,
+      "loss": 0.7918,
+      "step": 8645
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4470892514366442e-06,
+      "loss": 0.8085,
+      "step": 8646
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4454750967409737e-06,
+      "loss": 0.8165,
+      "step": 8647
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.443861772665044e-06,
+      "loss": 0.8151,
+      "step": 8648
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.442249279365512e-06,
+      "loss": 0.8399,
+      "step": 8649
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4406376169989389e-06,
+      "loss": 0.9506,
+      "step": 8650
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4390267857218188e-06,
+      "loss": 0.8749,
+      "step": 8651
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4374167856905542e-06,
+      "loss": 0.8694,
+      "step": 8652
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4358076170614754e-06,
+      "loss": 0.8913,
+      "step": 8653
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4341992799908255e-06,
+      "loss": 0.8092,
+      "step": 8654
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4325917746347672e-06,
+      "loss": 0.821,
+      "step": 8655
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4309851011493903e-06,
+      "loss": 0.8083,
+      "step": 8656
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.429379259690694e-06,
+      "loss": 0.8271,
+      "step": 8657
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.427774250414601e-06,
+      "loss": 0.8469,
+      "step": 8658
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4261700734769557e-06,
+      "loss": 0.8505,
+      "step": 8659
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4245667290335175e-06,
+      "loss": 0.948,
+      "step": 8660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4229642172399639e-06,
+      "loss": 0.853,
+      "step": 8661
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.421362538251897e-06,
+      "loss": 0.9167,
+      "step": 8662
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.419761692224837e-06,
+      "loss": 0.8448,
+      "step": 8663
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4181616793142173e-06,
+      "loss": 0.8611,
+      "step": 8664
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.416562499675398e-06,
+      "loss": 0.8181,
+      "step": 8665
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.414964153463655e-06,
+      "loss": 0.8723,
+      "step": 8666
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4133666408341784e-06,
+      "loss": 0.8685,
+      "step": 8667
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4117699619420878e-06,
+      "loss": 0.8509,
+      "step": 8668
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4101741169424133e-06,
+      "loss": 0.8031,
+      "step": 8669
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4085791059901077e-06,
+      "loss": 0.8722,
+      "step": 8670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4069849292400395e-06,
+      "loss": 0.8932,
+      "step": 8671
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4053915868470013e-06,
+      "loss": 0.824,
+      "step": 8672
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4037990789657008e-06,
+      "loss": 0.8319,
+      "step": 8673
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.402207405750765e-06,
+      "loss": 0.8334,
+      "step": 8674
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4006165673567462e-06,
+      "loss": 0.9051,
+      "step": 8675
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.399026563938105e-06,
+      "loss": 0.9472,
+      "step": 8676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.397437395649226e-06,
+      "loss": 0.8396,
+      "step": 8677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3958490626444154e-06,
+      "loss": 0.8071,
+      "step": 8678
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3942615650778956e-06,
+      "loss": 0.8823,
+      "step": 8679
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3926749031038055e-06,
+      "loss": 0.7946,
+      "step": 8680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3910890768762087e-06,
+      "loss": 0.8205,
+      "step": 8681
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3895040865490817e-06,
+      "loss": 0.8707,
+      "step": 8682
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3879199322763227e-06,
+      "loss": 0.8558,
+      "step": 8683
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3863366142117506e-06,
+      "loss": 0.835,
+      "step": 8684
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3847541325090985e-06,
+      "loss": 0.8004,
+      "step": 8685
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.383172487322023e-06,
+      "loss": 0.8373,
+      "step": 8686
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.381591678804095e-06,
+      "loss": 0.8709,
+      "step": 8687
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3800117071088104e-06,
+      "loss": 0.8219,
+      "step": 8688
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3784325723895765e-06,
+      "loss": 0.8896,
+      "step": 8689
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3768542747997215e-06,
+      "loss": 0.8768,
+      "step": 8690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3752768144924988e-06,
+      "loss": 0.8259,
+      "step": 8691
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3737001916210713e-06,
+      "loss": 0.8194,
+      "step": 8692
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.372124406338523e-06,
+      "loss": 0.7781,
+      "step": 8693
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.370549458797863e-06,
+      "loss": 0.8411,
+      "step": 8694
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3689753491520108e-06,
+      "loss": 0.8226,
+      "step": 8695
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3674020775538078e-06,
+      "loss": 0.8938,
+      "step": 8696
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3658296441560138e-06,
+      "loss": 0.8619,
+      "step": 8697
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3642580491113122e-06,
+      "loss": 0.8491,
+      "step": 8698
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3626872925722956e-06,
+      "loss": 0.8146,
+      "step": 8699
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3611173746914797e-06,
+      "loss": 0.8708,
+      "step": 8700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3595482956213036e-06,
+      "loss": 0.8683,
+      "step": 8701
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3579800555141165e-06,
+      "loss": 0.8875,
+      "step": 8702
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.356412654522189e-06,
+      "loss": 0.8987,
+      "step": 8703
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3548460927977158e-06,
+      "loss": 0.8108,
+      "step": 8704
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.353280370492802e-06,
+      "loss": 0.7798,
+      "step": 8705
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.351715487759474e-06,
+      "loss": 0.7731,
+      "step": 8706
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.35015144474968e-06,
+      "loss": 0.9266,
+      "step": 8707
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3485882416152819e-06,
+      "loss": 0.7889,
+      "step": 8708
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3470258785080659e-06,
+      "loss": 0.831,
+      "step": 8709
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3454643555797276e-06,
+      "loss": 0.8502,
+      "step": 8710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3439036729818911e-06,
+      "loss": 0.867,
+      "step": 8711
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3423438308660929e-06,
+      "loss": 0.8875,
+      "step": 8712
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3407848293837867e-06,
+      "loss": 0.8224,
+      "step": 8713
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3392266686863508e-06,
+      "loss": 0.8982,
+      "step": 8714
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.337669348925076e-06,
+      "loss": 0.834,
+      "step": 8715
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3361128702511716e-06,
+      "loss": 0.9268,
+      "step": 8716
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.334557232815772e-06,
+      "loss": 0.888,
+      "step": 8717
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3330024367699224e-06,
+      "loss": 0.8094,
+      "step": 8718
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3314484822645868e-06,
+      "loss": 0.8839,
+      "step": 8719
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3298953694506522e-06,
+      "loss": 0.7967,
+      "step": 8720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3283430984789236e-06,
+      "loss": 0.9261,
+      "step": 8721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3267916695001172e-06,
+      "loss": 0.912,
+      "step": 8722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3252410826648776e-06,
+      "loss": 0.839,
+      "step": 8723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3236913381237592e-06,
+      "loss": 0.8997,
+      "step": 8724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.322142436027235e-06,
+      "loss": 0.847,
+      "step": 8725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3205943765257057e-06,
+      "loss": 0.8124,
+      "step": 8726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3190471597694787e-06,
+      "loss": 0.7636,
+      "step": 8727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.317500785908783e-06,
+      "loss": 0.8268,
+      "step": 8728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3159552550937727e-06,
+      "loss": 0.884,
+      "step": 8729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.31441056747451e-06,
+      "loss": 0.9235,
+      "step": 8730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3128667232009806e-06,
+      "loss": 0.7654,
+      "step": 8731
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3113237224230836e-06,
+      "loss": 0.8284,
+      "step": 8732
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3097815652906476e-06,
+      "loss": 0.858,
+      "step": 8733
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3082402519534076e-06,
+      "loss": 0.9512,
+      "step": 8734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3066997825610195e-06,
+      "loss": 0.737,
+      "step": 8735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3051601572630611e-06,
+      "loss": 0.8433,
+      "step": 8736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3036213762090244e-06,
+      "loss": 0.8989,
+      "step": 8737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3020834395483195e-06,
+      "loss": 0.8645,
+      "step": 8738
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.300546347430277e-06,
+      "loss": 0.8849,
+      "step": 8739
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2990101000041445e-06,
+      "loss": 0.8112,
+      "step": 8740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2974746974190845e-06,
+      "loss": 0.7998,
+      "step": 8741
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2959401398241844e-06,
+      "loss": 0.807,
+      "step": 8742
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2944064273684409e-06,
+      "loss": 0.8706,
+      "step": 8743
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2928735602007768e-06,
+      "loss": 0.8546,
+      "step": 8744
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2913415384700257e-06,
+      "loss": 0.9209,
+      "step": 8745
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2898103623249458e-06,
+      "loss": 0.8747,
+      "step": 8746
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2882800319142085e-06,
+      "loss": 0.8627,
+      "step": 8747
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2867505473864029e-06,
+      "loss": 0.8345,
+      "step": 8748
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2852219088900409e-06,
+      "loss": 0.8411,
+      "step": 8749
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.283694116573546e-06,
+      "loss": 0.8539,
+      "step": 8750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2821671705852624e-06,
+      "loss": 0.8782,
+      "step": 8751
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2806410710734552e-06,
+      "loss": 0.8552,
+      "step": 8752
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2791158181863027e-06,
+      "loss": 0.8408,
+      "step": 8753
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2775914120718992e-06,
+      "loss": 0.8529,
+      "step": 8754
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2760678528782633e-06,
+      "loss": 0.9117,
+      "step": 8755
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2745451407533294e-06,
+      "loss": 0.8547,
+      "step": 8756
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.273023275844948e-06,
+      "loss": 0.9091,
+      "step": 8757
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2715022583008851e-06,
+      "loss": 0.8443,
+      "step": 8758
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2699820882688296e-06,
+      "loss": 0.8331,
+      "step": 8759
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2684627658963865e-06,
+      "loss": 0.9521,
+      "step": 8760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2669442913310725e-06,
+      "loss": 0.8645,
+      "step": 8761
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.265426664720334e-06,
+      "loss": 0.8841,
+      "step": 8762
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.263909886211524e-06,
+      "loss": 0.8984,
+      "step": 8763
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2623939559519161e-06,
+      "loss": 0.8568,
+      "step": 8764
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2608788740887069e-06,
+      "loss": 0.8716,
+      "step": 8765
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2593646407690051e-06,
+      "loss": 0.8358,
+      "step": 8766
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2578512561398348e-06,
+      "loss": 0.9207,
+      "step": 8767
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2563387203481447e-06,
+      "loss": 0.8106,
+      "step": 8768
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2548270335407985e-06,
+      "loss": 0.887,
+      "step": 8769
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2533161958645755e-06,
+      "loss": 0.8383,
+      "step": 8770
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2518062074661718e-06,
+      "loss": 0.8298,
+      "step": 8771
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2502970684922067e-06,
+      "loss": 0.8296,
+      "step": 8772
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2487887790892117e-06,
+      "loss": 0.8668,
+      "step": 8773
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2472813394036344e-06,
+      "loss": 0.8918,
+      "step": 8774
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2457747495818462e-06,
+      "loss": 0.9149,
+      "step": 8775
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2442690097701327e-06,
+      "loss": 0.907,
+      "step": 8776
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2427641201146945e-06,
+      "loss": 0.8043,
+      "step": 8777
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2412600807616526e-06,
+      "loss": 0.8402,
+      "step": 8778
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.239756891857048e-06,
+      "loss": 0.758,
+      "step": 8779
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2382545535468316e-06,
+      "loss": 0.898,
+      "step": 8780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.23675306597688e-06,
+      "loss": 0.7753,
+      "step": 8781
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2352524292929823e-06,
+      "loss": 0.8876,
+      "step": 8782
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.233752643640843e-06,
+      "loss": 0.8443,
+      "step": 8783
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2322537091660912e-06,
+      "loss": 0.8051,
+      "step": 8784
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2307556260142684e-06,
+      "loss": 0.8538,
+      "step": 8785
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.22925839433083e-06,
+      "loss": 0.8074,
+      "step": 8786
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.227762014261158e-06,
+      "loss": 0.9096,
+      "step": 8787
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2262664859505434e-06,
+      "loss": 0.8346,
+      "step": 8788
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2247718095441986e-06,
+      "loss": 0.8418,
+      "step": 8789
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2232779851872511e-06,
+      "loss": 0.8622,
+      "step": 8790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.221785013024751e-06,
+      "loss": 0.7598,
+      "step": 8791
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2202928932016588e-06,
+      "loss": 0.8763,
+      "step": 8792
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2188016258628531e-06,
+      "loss": 0.8289,
+      "step": 8793
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.217311211153137e-06,
+      "loss": 0.9064,
+      "step": 8794
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2158216492172215e-06,
+      "loss": 0.883,
+      "step": 8795
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2143329401997372e-06,
+      "loss": 0.8469,
+      "step": 8796
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2128450842452378e-06,
+      "loss": 0.7881,
+      "step": 8797
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2113580814981884e-06,
+      "loss": 0.8344,
+      "step": 8798
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2098719321029694e-06,
+      "loss": 0.8065,
+      "step": 8799
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2083866362038865e-06,
+      "loss": 0.8456,
+      "step": 8800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2069021939451565e-06,
+      "loss": 0.8752,
+      "step": 8801
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2054186054709105e-06,
+      "loss": 0.8012,
+      "step": 8802
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2039358709252036e-06,
+      "loss": 0.9142,
+      "step": 8803
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2024539904520072e-06,
+      "loss": 0.7817,
+      "step": 8804
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2009729641952061e-06,
+      "loss": 0.854,
+      "step": 8805
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1994927922985999e-06,
+      "loss": 0.8306,
+      "step": 8806
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1980134749059159e-06,
+      "loss": 0.7899,
+      "step": 8807
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1965350121607866e-06,
+      "loss": 0.7758,
+      "step": 8808
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1950574042067663e-06,
+      "loss": 0.8739,
+      "step": 8809
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1935806511873306e-06,
+      "loss": 0.87,
+      "step": 8810
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.192104753245864e-06,
+      "loss": 0.8685,
+      "step": 8811
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1906297105256725e-06,
+      "loss": 0.8414,
+      "step": 8812
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.189155523169978e-06,
+      "loss": 0.8416,
+      "step": 8813
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.187682191321925e-06,
+      "loss": 0.8135,
+      "step": 8814
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1862097151245644e-06,
+      "loss": 0.8916,
+      "step": 8815
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1847380947208697e-06,
+      "loss": 0.8654,
+      "step": 8816
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1832673302537334e-06,
+      "loss": 0.8537,
+      "step": 8817
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1817974218659621e-06,
+      "loss": 0.8409,
+      "step": 8818
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1803283697002765e-06,
+      "loss": 0.7693,
+      "step": 8819
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.178860173899321e-06,
+      "loss": 0.8438,
+      "step": 8820
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.177392834605653e-06,
+      "loss": 0.8561,
+      "step": 8821
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1759263519617437e-06,
+      "loss": 0.7804,
+      "step": 8822
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1744607261099883e-06,
+      "loss": 0.8309,
+      "step": 8823
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.172995957192693e-06,
+      "loss": 0.9097,
+      "step": 8824
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1715320453520807e-06,
+      "loss": 0.8453,
+      "step": 8825
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1700689907302953e-06,
+      "loss": 0.8638,
+      "step": 8826
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1686067934693967e-06,
+      "loss": 0.8121,
+      "step": 8827
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.167145453711358e-06,
+      "loss": 0.8412,
+      "step": 8828
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.165684971598069e-06,
+      "loss": 0.8932,
+      "step": 8829
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1642253472713427e-06,
+      "loss": 0.8488,
+      "step": 8830
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1627665808729039e-06,
+      "loss": 0.8479,
+      "step": 8831
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.161308672544389e-06,
+      "loss": 0.9009,
+      "step": 8832
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1598516224273637e-06,
+      "loss": 0.7587,
+      "step": 8833
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1583954306633004e-06,
+      "loss": 0.8975,
+      "step": 8834
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1569400973935895e-06,
+      "loss": 0.7949,
+      "step": 8835
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1554856227595435e-06,
+      "loss": 0.8696,
+      "step": 8836
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1540320069023825e-06,
+      "loss": 0.963,
+      "step": 8837
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1525792499632526e-06,
+      "loss": 0.8426,
+      "step": 8838
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1511273520832123e-06,
+      "loss": 0.8361,
+      "step": 8839
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1496763134032363e-06,
+      "loss": 0.8743,
+      "step": 8840
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1482261340642132e-06,
+      "loss": 0.8051,
+      "step": 8841
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1467768142069546e-06,
+      "loss": 0.811,
+      "step": 8842
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.145328353972186e-06,
+      "loss": 0.7997,
+      "step": 8843
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1438807535005437e-06,
+      "loss": 0.8555,
+      "step": 8844
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1424340129325917e-06,
+      "loss": 0.8021,
+      "step": 8845
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1409881324088013e-06,
+      "loss": 0.8495,
+      "step": 8846
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.139543112069561e-06,
+      "loss": 0.7509,
+      "step": 8847
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.138098952055181e-06,
+      "loss": 0.8041,
+      "step": 8848
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.136655652505887e-06,
+      "loss": 0.9194,
+      "step": 8849
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1352132135618165e-06,
+      "loss": 0.892,
+      "step": 8850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1337716353630258e-06,
+      "loss": 0.882,
+      "step": 8851
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.13233091804949e-06,
+      "loss": 0.7803,
+      "step": 8852
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.130891061761097e-06,
+      "loss": 0.8586,
+      "step": 8853
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1294520666376518e-06,
+      "loss": 0.8394,
+      "step": 8854
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1280139328188788e-06,
+      "loss": 0.8531,
+      "step": 8855
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1265766604444172e-06,
+      "loss": 0.8828,
+      "step": 8856
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1251402496538188e-06,
+      "loss": 0.8912,
+      "step": 8857
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1237047005865576e-06,
+      "loss": 0.8207,
+      "step": 8858
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1222700133820218e-06,
+      "loss": 0.7935,
+      "step": 8859
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1208361881795116e-06,
+      "loss": 0.8803,
+      "step": 8860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1194032251182496e-06,
+      "loss": 0.8413,
+      "step": 8861
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1179711243373736e-06,
+      "loss": 0.8943,
+      "step": 8862
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.116539885975937e-06,
+      "loss": 0.8679,
+      "step": 8863
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1151095101729047e-06,
+      "loss": 0.926,
+      "step": 8864
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1136799970671674e-06,
+      "loss": 0.7852,
+      "step": 8865
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1122513467975237e-06,
+      "loss": 0.8582,
+      "step": 8866
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1108235595026895e-06,
+      "loss": 0.8792,
+      "step": 8867
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1093966353213036e-06,
+      "loss": 0.8865,
+      "step": 8868
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1079705743919133e-06,
+      "loss": 0.8483,
+      "step": 8869
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1065453768529844e-06,
+      "loss": 0.8467,
+      "step": 8870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1051210428429017e-06,
+      "loss": 0.8914,
+      "step": 8871
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.103697572499961e-06,
+      "loss": 0.82,
+      "step": 8872
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.102274965962381e-06,
+      "loss": 0.7798,
+      "step": 8873
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1008532233682878e-06,
+      "loss": 0.9561,
+      "step": 8874
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0994323448557343e-06,
+      "loss": 0.8486,
+      "step": 8875
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0980123305626812e-06,
+      "loss": 0.791,
+      "step": 8876
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0965931806270048e-06,
+      "loss": 0.8415,
+      "step": 8877
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0951748951865048e-06,
+      "loss": 0.8468,
+      "step": 8878
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0937574743788915e-06,
+      "loss": 0.8149,
+      "step": 8879
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0923409183417887e-06,
+      "loss": 0.9441,
+      "step": 8880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0909252272127458e-06,
+      "loss": 0.86,
+      "step": 8881
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0895104011292202e-06,
+      "loss": 0.9258,
+      "step": 8882
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0880964402285843e-06,
+      "loss": 0.8599,
+      "step": 8883
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0866833446481317e-06,
+      "loss": 0.9062,
+      "step": 8884
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0852711145250715e-06,
+      "loss": 0.8509,
+      "step": 8885
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0838597499965276e-06,
+      "loss": 0.8116,
+      "step": 8886
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0824492511995354e-06,
+      "loss": 0.807,
+      "step": 8887
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0810396182710535e-06,
+      "loss": 0.8509,
+      "step": 8888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0796308513479537e-06,
+      "loss": 0.8316,
+      "step": 8889
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0782229505670195e-06,
+      "loss": 0.83,
+      "step": 8890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0768159160649593e-06,
+      "loss": 0.8278,
+      "step": 8891
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0754097479783876e-06,
+      "loss": 0.8802,
+      "step": 8892
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0740044464438404e-06,
+      "loss": 0.6955,
+      "step": 8893
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0726000115977696e-06,
+      "loss": 0.8596,
+      "step": 8894
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0711964435765399e-06,
+      "loss": 0.8199,
+      "step": 8895
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.069793742516435e-06,
+      "loss": 0.9054,
+      "step": 8896
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0683919085536553e-06,
+      "loss": 0.8466,
+      "step": 8897
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.066990941824312e-06,
+      "loss": 0.846,
+      "step": 8898
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0655908424644335e-06,
+      "loss": 0.867,
+      "step": 8899
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0641916106099691e-06,
+      "loss": 0.8992,
+      "step": 8900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0627932463967793e-06,
+      "loss": 0.8159,
+      "step": 8901
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0613957499606388e-06,
+      "loss": 0.8239,
+      "step": 8902
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.059999121437244e-06,
+      "loss": 0.8572,
+      "step": 8903
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0586033609622004e-06,
+      "loss": 0.8386,
+      "step": 8904
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0572084686710327e-06,
+      "loss": 0.8105,
+      "step": 8905
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0558144446991836e-06,
+      "loss": 0.8397,
+      "step": 8906
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.054421289182005e-06,
+      "loss": 0.8504,
+      "step": 8907
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.053029002254773e-06,
+      "loss": 0.8683,
+      "step": 8908
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0516375840526693e-06,
+      "loss": 0.8871,
+      "step": 8909
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0502470347108017e-06,
+      "loss": 0.8338,
+      "step": 8910
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0488573543641855e-06,
+      "loss": 0.9181,
+      "step": 8911
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0474685431477537e-06,
+      "loss": 0.8193,
+      "step": 8912
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0460806011963599e-06,
+      "loss": 0.8348,
+      "step": 8913
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0446935286447657e-06,
+      "loss": 0.8177,
+      "step": 8914
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0433073256276517e-06,
+      "loss": 0.7999,
+      "step": 8915
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0419219922796175e-06,
+      "loss": 0.7919,
+      "step": 8916
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0405375287351727e-06,
+      "loss": 0.8704,
+      "step": 8917
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.039153935128744e-06,
+      "loss": 0.9366,
+      "step": 8918
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0377712115946736e-06,
+      "loss": 0.8441,
+      "step": 8919
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0363893582672246e-06,
+      "loss": 0.8868,
+      "step": 8920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0350083752805685e-06,
+      "loss": 0.8183,
+      "step": 8921
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.033628262768792e-06,
+      "loss": 0.8781,
+      "step": 8922
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0322490208659053e-06,
+      "loss": 0.8758,
+      "step": 8923
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0308706497058252e-06,
+      "loss": 0.8687,
+      "step": 8924
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0294931494223858e-06,
+      "loss": 0.8058,
+      "step": 8925
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0281165201493437e-06,
+      "loss": 0.8398,
+      "step": 8926
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0267407620203628e-06,
+      "loss": 0.8588,
+      "step": 8927
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0253658751690232e-06,
+      "loss": 0.8257,
+      "step": 8928
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0239918597288268e-06,
+      "loss": 0.8758,
+      "step": 8929
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0226187158331825e-06,
+      "loss": 0.8848,
+      "step": 8930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0212464436154225e-06,
+      "loss": 0.8159,
+      "step": 8931
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0198750432087855e-06,
+      "loss": 0.8459,
+      "step": 8932
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0185045147464356e-06,
+      "loss": 0.8508,
+      "step": 8933
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.017134858361446e-06,
+      "loss": 0.8065,
+      "step": 8934
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0157660741868036e-06,
+      "loss": 0.8704,
+      "step": 8935
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0143981623554155e-06,
+      "loss": 0.8372,
+      "step": 8936
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0130311230001033e-06,
+      "loss": 0.8029,
+      "step": 8937
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0116649562535984e-06,
+      "loss": 0.8703,
+      "step": 8938
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0102996622485573e-06,
+      "loss": 0.8685,
+      "step": 8939
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0089352411175424e-06,
+      "loss": 0.8375,
+      "step": 8940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0075716929930334e-06,
+      "loss": 0.9259,
+      "step": 8941
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.006209018007429e-06,
+      "loss": 0.9097,
+      "step": 8942
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0048472162930445e-06,
+      "loss": 0.8244,
+      "step": 8943
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0034862879821029e-06,
+      "loss": 0.871,
+      "step": 8944
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.002126233206745e-06,
+      "loss": 0.8741,
+      "step": 8945
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0007670520990331e-06,
+      "loss": 0.862,
+      "step": 8946
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.994087447909372e-07,
+      "loss": 0.8481,
+      "step": 8947
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98051311414342e-07,
+      "loss": 0.8095,
+      "step": 8948
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.96694752101056e-07,
+      "loss": 0.8284,
+      "step": 8949
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.953390669827944e-07,
+      "loss": 0.8579,
+      "step": 8950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.939842561911895e-07,
+      "loss": 0.9395,
+      "step": 8951
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.926303198577913e-07,
+      "loss": 0.8407,
+      "step": 8952
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.912772581140606e-07,
+      "loss": 0.8828,
+      "step": 8953
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.899250710913767e-07,
+      "loss": 0.944,
+      "step": 8954
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.885737589210366e-07,
+      "loss": 0.8099,
+      "step": 8955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.872233217342463e-07,
+      "loss": 0.8356,
+      "step": 8956
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.85873759662127e-07,
+      "loss": 0.8774,
+      "step": 8957
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.845250728357214e-07,
+      "loss": 0.8666,
+      "step": 8958
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.831772613859814e-07,
+      "loss": 0.9078,
+      "step": 8959
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.818303254437723e-07,
+      "loss": 0.8601,
+      "step": 8960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.804842651398839e-07,
+      "loss": 0.8147,
+      "step": 8961
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.791390806050117e-07,
+      "loss": 0.8199,
+      "step": 8962
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.777947719697667e-07,
+      "loss": 0.8601,
+      "step": 8963
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.764513393646812e-07,
+      "loss": 0.8409,
+      "step": 8964
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.751087829201966e-07,
+      "loss": 0.8633,
+      "step": 8965
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.737671027666728e-07,
+      "loss": 0.8784,
+      "step": 8966
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.724262990343813e-07,
+      "loss": 0.891,
+      "step": 8967
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.710863718535135e-07,
+      "loss": 0.8701,
+      "step": 8968
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.697473213541697e-07,
+      "loss": 0.8593,
+      "step": 8969
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.684091476663659e-07,
+      "loss": 0.8505,
+      "step": 8970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.670718509200393e-07,
+      "loss": 0.8807,
+      "step": 8971
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.657354312450363e-07,
+      "loss": 0.8036,
+      "step": 8972
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.64399888771117e-07,
+      "loss": 0.8365,
+      "step": 8973
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.630652236279626e-07,
+      "loss": 0.8343,
+      "step": 8974
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.617314359451624e-07,
+      "loss": 0.8475,
+      "step": 8975
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.603985258522219e-07,
+      "loss": 0.8617,
+      "step": 8976
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.590664934785666e-07,
+      "loss": 0.8183,
+      "step": 8977
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.577353389535315e-07,
+      "loss": 0.8587,
+      "step": 8978
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.564050624063691e-07,
+      "loss": 0.8409,
+      "step": 8979
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.550756639662417e-07,
+      "loss": 0.8245,
+      "step": 8980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.537471437622347e-07,
+      "loss": 0.8196,
+      "step": 8981
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.524195019233407e-07,
+      "loss": 0.8642,
+      "step": 8982
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.510927385784686e-07,
+      "loss": 0.7943,
+      "step": 8983
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.497668538564475e-07,
+      "loss": 0.8014,
+      "step": 8984
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.484418478860158e-07,
+      "loss": 0.8398,
+      "step": 8985
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.471177207958238e-07,
+      "loss": 0.8646,
+      "step": 8986
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.457944727144453e-07,
+      "loss": 0.7788,
+      "step": 8987
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.444721037703597e-07,
+      "loss": 0.8397,
+      "step": 8988
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.431506140919699e-07,
+      "loss": 0.7996,
+      "step": 8989
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.418300038075845e-07,
+      "loss": 0.8311,
+      "step": 8990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.40510273045434e-07,
+      "loss": 0.8307,
+      "step": 8991
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.391914219336606e-07,
+      "loss": 0.8092,
+      "step": 8992
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.378734506003162e-07,
+      "loss": 0.9072,
+      "step": 8993
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.365563591733784e-07,
+      "loss": 0.8621,
+      "step": 8994
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.352401477807293e-07,
+      "loss": 0.8467,
+      "step": 8995
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.33924816550168e-07,
+      "loss": 0.8309,
+      "step": 8996
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.326103656094132e-07,
+      "loss": 0.8051,
+      "step": 8997
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.31296795086093e-07,
+      "loss": 0.8602,
+      "step": 8998
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.299841051077474e-07,
+      "loss": 0.8694,
+      "step": 8999
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.286722958018391e-07,
+      "loss": 0.8788,
+      "step": 9000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.273613672957427e-07,
+      "loss": 0.7993,
+      "step": 9001
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.260513197167398e-07,
+      "loss": 0.8884,
+      "step": 9002
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.247421531920375e-07,
+      "loss": 0.7771,
+      "step": 9003
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.234338678487509e-07,
+      "loss": 0.8293,
+      "step": 9004
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.221264638139082e-07,
+      "loss": 0.8951,
+      "step": 9005
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.208199412144559e-07,
+      "loss": 0.8624,
+      "step": 9006
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.195143001772544e-07,
+      "loss": 0.9053,
+      "step": 9007
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.182095408290781e-07,
+      "loss": 0.8176,
+      "step": 9008
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.169056632966123e-07,
+      "loss": 0.8098,
+      "step": 9009
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.156026677064633e-07,
+      "loss": 0.8811,
+      "step": 9010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.143005541851469e-07,
+      "loss": 0.8371,
+      "step": 9011
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.129993228590917e-07,
+      "loss": 0.9037,
+      "step": 9012
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.116989738546489e-07,
+      "loss": 0.8458,
+      "step": 9013
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.103995072980765e-07,
+      "loss": 0.8581,
+      "step": 9014
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.091009233155467e-07,
+      "loss": 0.8381,
+      "step": 9015
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.078032220331523e-07,
+      "loss": 0.8473,
+      "step": 9016
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.065064035768933e-07,
+      "loss": 0.9001,
+      "step": 9017
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.052104680726859e-07,
+      "loss": 0.8735,
+      "step": 9018
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.03915415646367e-07,
+      "loss": 0.8666,
+      "step": 9019
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.026212464236772e-07,
+      "loss": 0.7979,
+      "step": 9020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.013279605302782e-07,
+      "loss": 0.8679,
+      "step": 9021
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.000355580917464e-07,
+      "loss": 0.7868,
+      "step": 9022
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.987440392335656e-07,
+      "loss": 0.8406,
+      "step": 9023
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.974534040811444e-07,
+      "loss": 0.8398,
+      "step": 9024
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.96163652759795e-07,
+      "loss": 0.8373,
+      "step": 9025
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.948747853947526e-07,
+      "loss": 0.9356,
+      "step": 9026
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.935868021111593e-07,
+      "loss": 0.8293,
+      "step": 9027
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.922997030340752e-07,
+      "loss": 0.7926,
+      "step": 9028
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.910134882884757e-07,
+      "loss": 0.8379,
+      "step": 9029
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.897281579992467e-07,
+      "loss": 0.8357,
+      "step": 9030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.884437122911904e-07,
+      "loss": 0.8271,
+      "step": 9031
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.871601512890238e-07,
+      "loss": 0.7647,
+      "step": 9032
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.85877475117376e-07,
+      "loss": 0.8206,
+      "step": 9033
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.845956839007897e-07,
+      "loss": 0.8734,
+      "step": 9034
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.833147777637241e-07,
+      "loss": 0.807,
+      "step": 9035
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.820347568305543e-07,
+      "loss": 0.8707,
+      "step": 9036
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.807556212255652e-07,
+      "loss": 0.8684,
+      "step": 9037
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.794773710729543e-07,
+      "loss": 0.8371,
+      "step": 9038
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.782000064968399e-07,
+      "loss": 0.8262,
+      "step": 9039
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.769235276212496e-07,
+      "loss": 0.8716,
+      "step": 9040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.756479345701218e-07,
+      "loss": 0.8553,
+      "step": 9041
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.743732274673189e-07,
+      "loss": 0.8803,
+      "step": 9042
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.730994064366083e-07,
+      "loss": 0.8451,
+      "step": 9043
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.718264716016722e-07,
+      "loss": 0.8743,
+      "step": 9044
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.705544230861141e-07,
+      "loss": 0.7544,
+      "step": 9045
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.692832610134428e-07,
+      "loss": 0.8672,
+      "step": 9046
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.680129855070829e-07,
+      "loss": 0.8316,
+      "step": 9047
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.66743596690377e-07,
+      "loss": 0.8453,
+      "step": 9048
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.654750946865809e-07,
+      "loss": 0.9062,
+      "step": 9049
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.642074796188594e-07,
+      "loss": 0.8859,
+      "step": 9050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.629407516102939e-07,
+      "loss": 0.8554,
+      "step": 9051
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.61674910783884e-07,
+      "loss": 0.9366,
+      "step": 9052
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.604099572625357e-07,
+      "loss": 0.8651,
+      "step": 9053
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.59145891169072e-07,
+      "loss": 0.8692,
+      "step": 9054
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.578827126262324e-07,
+      "loss": 0.9039,
+      "step": 9055
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.566204217566664e-07,
+      "loss": 0.927,
+      "step": 9056
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.553590186829386e-07,
+      "loss": 0.8273,
+      "step": 9057
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.540985035275273e-07,
+      "loss": 0.8779,
+      "step": 9058
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.528388764128282e-07,
+      "loss": 0.9074,
+      "step": 9059
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.515801374611432e-07,
+      "loss": 0.8584,
+      "step": 9060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.503222867946958e-07,
+      "loss": 0.8652,
+      "step": 9061
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.490653245356184e-07,
+      "loss": 0.8657,
+      "step": 9062
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.478092508059554e-07,
+      "loss": 0.8215,
+      "step": 9063
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.465540657276728e-07,
+      "loss": 0.8462,
+      "step": 9064
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.452997694226439e-07,
+      "loss": 0.7113,
+      "step": 9065
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.44046362012656e-07,
+      "loss": 0.7942,
+      "step": 9066
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.427938436194094e-07,
+      "loss": 0.8477,
+      "step": 9067
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.415422143645247e-07,
+      "loss": 0.7973,
+      "step": 9068
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.402914743695301e-07,
+      "loss": 0.793,
+      "step": 9069
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.390416237558641e-07,
+      "loss": 0.889,
+      "step": 9070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.377926626448907e-07,
+      "loss": 0.8305,
+      "step": 9071
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.365445911578785e-07,
+      "loss": 0.8248,
+      "step": 9072
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.352974094160071e-07,
+      "loss": 0.7892,
+      "step": 9073
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.340511175403809e-07,
+      "loss": 0.8492,
+      "step": 9074
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.328057156520064e-07,
+      "loss": 0.9207,
+      "step": 9075
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.315612038718101e-07,
+      "loss": 0.8,
+      "step": 9076
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.30317582320631e-07,
+      "loss": 0.9373,
+      "step": 9077
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.290748511192214e-07,
+      "loss": 0.8662,
+      "step": 9078
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.278330103882437e-07,
+      "loss": 0.7775,
+      "step": 9079
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.265920602482825e-07,
+      "loss": 0.8326,
+      "step": 9080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.253520008198268e-07,
+      "loss": 0.8481,
+      "step": 9081
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.241128322232816e-07,
+      "loss": 0.8835,
+      "step": 9082
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.228745545789696e-07,
+      "loss": 0.9016,
+      "step": 9083
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.216371680071244e-07,
+      "loss": 0.8897,
+      "step": 9084
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.204006726278902e-07,
+      "loss": 0.9196,
+      "step": 9085
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.191650685613273e-07,
+      "loss": 0.8854,
+      "step": 9086
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.179303559274121e-07,
+      "loss": 0.898,
+      "step": 9087
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.166965348460298e-07,
+      "loss": 0.7903,
+      "step": 9088
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.15463605436978e-07,
+      "loss": 0.8297,
+      "step": 9089
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.142315678199764e-07,
+      "loss": 0.8099,
+      "step": 9090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.130004221146492e-07,
+      "loss": 0.9135,
+      "step": 9091
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.117701684405343e-07,
+      "loss": 0.8939,
+      "step": 9092
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.105408069170906e-07,
+      "loss": 0.8683,
+      "step": 9093
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.093123376636836e-07,
+      "loss": 0.8622,
+      "step": 9094
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.080847607995956e-07,
+      "loss": 0.9029,
+      "step": 9095
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.06858076444017e-07,
+      "loss": 0.905,
+      "step": 9096
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.056322847160592e-07,
+      "loss": 0.925,
+      "step": 9097
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.044073857347423e-07,
+      "loss": 0.9325,
+      "step": 9098
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.031833796189969e-07,
+      "loss": 0.8618,
+      "step": 9099
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.019602664876758e-07,
+      "loss": 0.8566,
+      "step": 9100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.007380464595371e-07,
+      "loss": 0.8646,
+      "step": 9101
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.995167196532527e-07,
+      "loss": 0.8755,
+      "step": 9102
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.982962861874155e-07,
+      "loss": 0.8663,
+      "step": 9103
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.970767461805218e-07,
+      "loss": 0.9464,
+      "step": 9104
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.958580997509846e-07,
+      "loss": 0.8462,
+      "step": 9105
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.946403470171326e-07,
+      "loss": 0.8596,
+      "step": 9106
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.93423488097208e-07,
+      "loss": 0.8258,
+      "step": 9107
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.922075231093628e-07,
+      "loss": 0.8885,
+      "step": 9108
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.909924521716617e-07,
+      "loss": 0.8754,
+      "step": 9109
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.897782754020889e-07,
+      "loss": 0.8577,
+      "step": 9110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.885649929185335e-07,
+      "loss": 0.79,
+      "step": 9111
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.873526048388025e-07,
+      "loss": 0.7816,
+      "step": 9112
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.861411112806172e-07,
+      "loss": 0.8667,
+      "step": 9113
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.849305123616091e-07,
+      "loss": 0.8667,
+      "step": 9114
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83720808199322e-07,
+      "loss": 0.8906,
+      "step": 9115
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.825119989112173e-07,
+      "loss": 0.8152,
+      "step": 9116
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.813040846146646e-07,
+      "loss": 0.9248,
+      "step": 9117
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800970654269513e-07,
+      "loss": 0.8169,
+      "step": 9118
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.788909414652746e-07,
+      "loss": 0.8252,
+      "step": 9119
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.776857128467464e-07,
+      "loss": 0.8253,
+      "step": 9120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.764813796883885e-07,
+      "loss": 0.9106,
+      "step": 9121
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75277942107141e-07,
+      "loss": 0.763,
+      "step": 9122
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.740754002198536e-07,
+      "loss": 0.8945,
+      "step": 9123
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.728737541432862e-07,
+      "loss": 0.9191,
+      "step": 9124
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716730039941201e-07,
+      "loss": 0.8461,
+      "step": 9125
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.704731498889428e-07,
+      "loss": 0.8642,
+      "step": 9126
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69274191944257e-07,
+      "loss": 0.8848,
+      "step": 9127
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680761302764727e-07,
+      "loss": 0.8026,
+      "step": 9128
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.668789650019281e-07,
+      "loss": 0.8918,
+      "step": 9129
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65682696236858e-07,
+      "loss": 0.8384,
+      "step": 9130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.644873240974171e-07,
+      "loss": 0.787,
+      "step": 9131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.632928486996749e-07,
+      "loss": 0.8174,
+      "step": 9132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.620992701596108e-07,
+      "loss": 0.8784,
+      "step": 9133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.609065885931155e-07,
+      "loss": 0.8395,
+      "step": 9134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.597148041159985e-07,
+      "loss": 0.8568,
+      "step": 9135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.585239168439762e-07,
+      "loss": 0.8004,
+      "step": 9136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.573339268926794e-07,
+      "loss": 0.8009,
+      "step": 9137
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.561448343776567e-07,
+      "loss": 0.8608,
+      "step": 9138
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.549566394143626e-07,
+      "loss": 0.8452,
+      "step": 9139
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.537693421181658e-07,
+      "loss": 0.757,
+      "step": 9140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.52582942604353e-07,
+      "loss": 0.9012,
+      "step": 9141
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.513974409881186e-07,
+      "loss": 0.81,
+      "step": 9142
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.50212837384573e-07,
+      "loss": 0.8693,
+      "step": 9143
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.49029131908734e-07,
+      "loss": 0.868,
+      "step": 9144
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.478463246755407e-07,
+      "loss": 0.8936,
+      "step": 9145
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.466644157998371e-07,
+      "loss": 0.7624,
+      "step": 9146
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.454834053963833e-07,
+      "loss": 0.8085,
+      "step": 9147
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.443032935798533e-07,
+      "loss": 0.8637,
+      "step": 9148
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.431240804648321e-07,
+      "loss": 0.8998,
+      "step": 9149
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.419457661658169e-07,
+      "loss": 0.8806,
+      "step": 9150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.407683507972208e-07,
+      "loss": 0.8257,
+      "step": 9151
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.395918344733644e-07,
+      "loss": 0.7808,
+      "step": 9152
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.384162173084874e-07,
+      "loss": 0.8328,
+      "step": 9153
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.372414994167354e-07,
+      "loss": 0.8641,
+      "step": 9154
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.360676809121736e-07,
+      "loss": 0.8417,
+      "step": 9155
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.348947619087754e-07,
+      "loss": 0.783,
+      "step": 9156
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.337227425204252e-07,
+      "loss": 0.8644,
+      "step": 9157
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.325516228609264e-07,
+      "loss": 0.867,
+      "step": 9158
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.313814030439892e-07,
+      "loss": 0.8793,
+      "step": 9159
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.302120831832382e-07,
+      "loss": 0.8447,
+      "step": 9160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.290436633922127e-07,
+      "loss": 0.9294,
+      "step": 9161
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.278761437843629e-07,
+      "loss": 0.8457,
+      "step": 9162
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.267095244730482e-07,
+      "loss": 0.7433,
+      "step": 9163
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.255438055715469e-07,
+      "loss": 0.8388,
+      "step": 9164
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.243789871930485e-07,
+      "loss": 0.84,
+      "step": 9165
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.232150694506512e-07,
+      "loss": 0.8154,
+      "step": 9166
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.22052052457366e-07,
+      "loss": 0.8819,
+      "step": 9167
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.208899363261234e-07,
+      "loss": 0.8498,
+      "step": 9168
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.19728721169759e-07,
+      "loss": 0.855,
+      "step": 9169
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.185684071010224e-07,
+      "loss": 0.8438,
+      "step": 9170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.174089942325791e-07,
+      "loss": 0.8224,
+      "step": 9171
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.162504826770033e-07,
+      "loss": 0.9023,
+      "step": 9172
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.15092872546782e-07,
+      "loss": 0.884,
+      "step": 9173
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.139361639543185e-07,
+      "loss": 0.8143,
+      "step": 9174
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.127803570119229e-07,
+      "loss": 0.8921,
+      "step": 9175
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.116254518318222e-07,
+      "loss": 0.8759,
+      "step": 9176
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.104714485261565e-07,
+      "loss": 0.8885,
+      "step": 9177
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.093183472069753e-07,
+      "loss": 0.9105,
+      "step": 9178
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.081661479862367e-07,
+      "loss": 0.872,
+      "step": 9179
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.070148509758223e-07,
+      "loss": 0.9294,
+      "step": 9180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.058644562875172e-07,
+      "loss": 0.8409,
+      "step": 9181
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.047149640330197e-07,
+      "loss": 0.8262,
+      "step": 9182
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.035663743239451e-07,
+      "loss": 0.8466,
+      "step": 9183
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.024186872718164e-07,
+      "loss": 0.8634,
+      "step": 9184
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.012719029880699e-07,
+      "loss": 0.8517,
+      "step": 9185
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.001260215840567e-07,
+      "loss": 0.8625,
+      "step": 9186
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.989810431710375e-07,
+      "loss": 0.8577,
+      "step": 9187
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.978369678601892e-07,
+      "loss": 0.8577,
+      "step": 9188
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.96693795762593e-07,
+      "loss": 0.9121,
+      "step": 9189
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.955515269892533e-07,
+      "loss": 0.9293,
+      "step": 9190
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.94410161651079e-07,
+      "loss": 0.8411,
+      "step": 9191
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.932696998588895e-07,
+      "loss": 0.7971,
+      "step": 9192
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.921301417234272e-07,
+      "loss": 0.8365,
+      "step": 9193
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.909914873553347e-07,
+      "loss": 0.8649,
+      "step": 9194
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.898537368651737e-07,
+      "loss": 0.8018,
+      "step": 9195
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.887168903634178e-07,
+      "loss": 0.7842,
+      "step": 9196
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.8758094796045e-07,
+      "loss": 0.8224,
+      "step": 9197
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.864459097665654e-07,
+      "loss": 0.9211,
+      "step": 9198
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.853117758919758e-07,
+      "loss": 0.8655,
+      "step": 9199
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.84178546446802e-07,
+      "loss": 0.8484,
+      "step": 9200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.830462215410772e-07,
+      "loss": 0.8223,
+      "step": 9201
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.819148012847454e-07,
+      "loss": 0.9331,
+      "step": 9202
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.807842857876656e-07,
+      "loss": 0.7837,
+      "step": 9203
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.796546751596089e-07,
+      "loss": 0.8699,
+      "step": 9204
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.785259695102531e-07,
+      "loss": 0.845,
+      "step": 9205
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.77398168949196e-07,
+      "loss": 0.848,
+      "step": 9206
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.762712735859434e-07,
+      "loss": 0.8783,
+      "step": 9207
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.751452835299111e-07,
+      "loss": 0.8,
+      "step": 9208
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.740201988904327e-07,
+      "loss": 0.8679,
+      "step": 9209
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.728960197767475e-07,
+      "loss": 0.854,
+      "step": 9210
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.717727462980128e-07,
+      "loss": 0.8055,
+      "step": 9211
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.706503785632934e-07,
+      "loss": 0.7949,
+      "step": 9212
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.695289166815699e-07,
+      "loss": 0.7841,
+      "step": 9213
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.68408360761732e-07,
+      "loss": 0.849,
+      "step": 9214
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.672887109125803e-07,
+      "loss": 0.8212,
+      "step": 9215
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.661699672428334e-07,
+      "loss": 0.9107,
+      "step": 9216
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.650521298611157e-07,
+      "loss": 0.8916,
+      "step": 9217
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.639351988759657e-07,
+      "loss": 0.7722,
+      "step": 9218
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.628191743958367e-07,
+      "loss": 0.8925,
+      "step": 9219
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.6170405652909e-07,
+      "loss": 0.8893,
+      "step": 9220
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.605898453839976e-07,
+      "loss": 0.864,
+      "step": 9221
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.594765410687487e-07,
+      "loss": 0.8846,
+      "step": 9222
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.583641436914445e-07,
+      "loss": 0.8696,
+      "step": 9223
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.57252653360092e-07,
+      "loss": 0.8819,
+      "step": 9224
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.561420701826149e-07,
+      "loss": 0.8155,
+      "step": 9225
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.550323942668469e-07,
+      "loss": 0.7729,
+      "step": 9226
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.539236257205361e-07,
+      "loss": 0.9058,
+      "step": 9227
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.528157646513378e-07,
+      "loss": 0.8522,
+      "step": 9228
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.517088111668257e-07,
+      "loss": 0.8192,
+      "step": 9229
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.506027653744796e-07,
+      "loss": 0.8277,
+      "step": 9230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.494976273816933e-07,
+      "loss": 0.7128,
+      "step": 9231
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.483933972957734e-07,
+      "loss": 0.9261,
+      "step": 9232
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.472900752239353e-07,
+      "loss": 0.871,
+      "step": 9233
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.461876612733109e-07,
+      "loss": 0.9037,
+      "step": 9234
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.450861555509413e-07,
+      "loss": 0.8337,
+      "step": 9235
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.4398555816378e-07,
+      "loss": 0.8095,
+      "step": 9236
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.428858692186879e-07,
+      "loss": 0.8257,
+      "step": 9237
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.417870888224476e-07,
+      "loss": 0.8806,
+      "step": 9238
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.406892170817425e-07,
+      "loss": 0.9067,
+      "step": 9239
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.395922541031741e-07,
+      "loss": 0.8365,
+      "step": 9240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.384961999932571e-07,
+      "loss": 0.815,
+      "step": 9241
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.374010548584119e-07,
+      "loss": 0.7809,
+      "step": 9242
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.363068188049736e-07,
+      "loss": 0.7929,
+      "step": 9243
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.352134919391928e-07,
+      "loss": 0.8379,
+      "step": 9244
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.341210743672244e-07,
+      "loss": 0.8508,
+      "step": 9245
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.330295661951436e-07,
+      "loss": 0.9088,
+      "step": 9246
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.319389675289278e-07,
+      "loss": 0.8475,
+      "step": 9247
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.308492784744746e-07,
+      "loss": 0.8438,
+      "step": 9248
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.297604991375883e-07,
+      "loss": 0.9306,
+      "step": 9249
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.286726296239854e-07,
+      "loss": 0.8957,
+      "step": 9250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.275856700392969e-07,
+      "loss": 0.8896,
+      "step": 9251
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.264996204890628e-07,
+      "loss": 0.8117,
+      "step": 9252
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.254144810787321e-07,
+      "loss": 0.9138,
+      "step": 9253
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.24330251913674e-07,
+      "loss": 0.9323,
+      "step": 9254
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.23246933099162e-07,
+      "loss": 0.8179,
+      "step": 9255
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.221645247403807e-07,
+      "loss": 0.8392,
+      "step": 9256
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.210830269424317e-07,
+      "loss": 0.8759,
+      "step": 9257
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.200024398103255e-07,
+      "loss": 0.8543,
+      "step": 9258
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.189227634489836e-07,
+      "loss": 0.8589,
+      "step": 9259
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.178439979632367e-07,
+      "loss": 0.8121,
+      "step": 9260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.167661434578331e-07,
+      "loss": 0.8266,
+      "step": 9261
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.156892000374293e-07,
+      "loss": 0.8583,
+      "step": 9262
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.146131678065914e-07,
+      "loss": 0.8884,
+      "step": 9263
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.135380468698004e-07,
+      "loss": 0.9139,
+      "step": 9264
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.124638373314484e-07,
+      "loss": 0.8775,
+      "step": 9265
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.113905392958342e-07,
+      "loss": 0.8153,
+      "step": 9266
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.103181528671764e-07,
+      "loss": 0.7766,
+      "step": 9267
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.092466781495976e-07,
+      "loss": 0.7893,
+      "step": 9268
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.081761152471366e-07,
+      "loss": 0.9173,
+      "step": 9269
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.071064642637404e-07,
+      "loss": 0.8849,
+      "step": 9270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.060377253032724e-07,
+      "loss": 0.9056,
+      "step": 9271
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.049698984695007e-07,
+      "loss": 0.7289,
+      "step": 9272
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.039029838661081e-07,
+      "loss": 0.8571,
+      "step": 9273
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.028369815966917e-07,
+      "loss": 0.8637,
+      "step": 9274
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.017718917647564e-07,
+      "loss": 0.8374,
+      "step": 9275
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.007077144737161e-07,
+      "loss": 0.8598,
+      "step": 9276
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.996444498269039e-07,
+      "loss": 0.8765,
+      "step": 9277
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.985820979275569e-07,
+      "loss": 0.7723,
+      "step": 9278
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.975206588788263e-07,
+      "loss": 0.8817,
+      "step": 9279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.96460132783776e-07,
+      "loss": 0.8098,
+      "step": 9280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.954005197453804e-07,
+      "loss": 0.8954,
+      "step": 9281
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.943418198665251e-07,
+      "loss": 0.9178,
+      "step": 9282
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.932840332500034e-07,
+      "loss": 0.7951,
+      "step": 9283
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.922271599985286e-07,
+      "loss": 0.7738,
+      "step": 9284
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.911712002147163e-07,
+      "loss": 0.7586,
+      "step": 9285
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.901161540010969e-07,
+      "loss": 0.9063,
+      "step": 9286
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.890620214601161e-07,
+      "loss": 0.8693,
+      "step": 9287
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.880088026941233e-07,
+      "loss": 0.8461,
+      "step": 9288
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.869564978053832e-07,
+      "loss": 0.8451,
+      "step": 9289
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.859051068960741e-07,
+      "loss": 0.8458,
+      "step": 9290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.848546300682823e-07,
+      "loss": 0.8545,
+      "step": 9291
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.838050674240025e-07,
+      "loss": 0.7944,
+      "step": 9292
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.827564190651492e-07,
+      "loss": 0.8369,
+      "step": 9293
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.817086850935416e-07,
+      "loss": 0.8777,
+      "step": 9294
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.806618656109098e-07,
+      "loss": 0.8636,
+      "step": 9295
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.796159607189001e-07,
+      "loss": 0.8154,
+      "step": 9296
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785709705190645e-07,
+      "loss": 0.8861,
+      "step": 9297
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.775268951128676e-07,
+      "loss": 0.7832,
+      "step": 9298
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.764837346016894e-07,
+      "loss": 0.6542,
+      "step": 9299
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.754414890868154e-07,
+      "loss": 0.9111,
+      "step": 9300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.744001586694437e-07,
+      "loss": 0.7961,
+      "step": 9301
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.733597434506877e-07,
+      "loss": 0.8435,
+      "step": 9302
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.723202435315667e-07,
+      "loss": 0.8199,
+      "step": 9303
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.712816590130133e-07,
+      "loss": 0.8574,
+      "step": 9304
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.702439899958711e-07,
+      "loss": 0.8546,
+      "step": 9305
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.692072365808954e-07,
+      "loss": 0.8975,
+      "step": 9306
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.681713988687521e-07,
+      "loss": 0.7907,
+      "step": 9307
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.671364769600162e-07,
+      "loss": 0.7677,
+      "step": 9308
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.661024709551777e-07,
+      "loss": 0.8474,
+      "step": 9309
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.650693809546348e-07,
+      "loss": 0.8831,
+      "step": 9310
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.640372070586963e-07,
+      "loss": 0.8933,
+      "step": 9311
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.630059493675866e-07,
+      "loss": 0.8677,
+      "step": 9312
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.619756079814354e-07,
+      "loss": 0.847,
+      "step": 9313
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.60946183000285e-07,
+      "loss": 0.8128,
+      "step": 9314
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.599176745240908e-07,
+      "loss": 0.8222,
+      "step": 9315
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.5889008265272e-07,
+      "loss": 0.891,
+      "step": 9316
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.578634074859457e-07,
+      "loss": 0.9349,
+      "step": 9317
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.568376491234562e-07,
+      "loss": 0.8166,
+      "step": 9318
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.558128076648506e-07,
+      "loss": 0.8721,
+      "step": 9319
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.547888832096382e-07,
+      "loss": 0.8271,
+      "step": 9320
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.537658758572362e-07,
+      "loss": 0.8707,
+      "step": 9321
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.527437857069784e-07,
+      "loss": 0.8039,
+      "step": 9322
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.517226128581065e-07,
+      "loss": 0.7298,
+      "step": 9323
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.507023574097725e-07,
+      "loss": 0.8434,
+      "step": 9324
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.496830194610414e-07,
+      "loss": 0.9034,
+      "step": 9325
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.486645991108875e-07,
+      "loss": 0.8191,
+      "step": 9326
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.476470964581948e-07,
+      "loss": 0.8706,
+      "step": 9327
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.466305116017623e-07,
+      "loss": 0.9124,
+      "step": 9328
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.456148446402976e-07,
+      "loss": 0.886,
+      "step": 9329
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.446000956724174e-07,
+      "loss": 0.7442,
+      "step": 9330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.435862647966506e-07,
+      "loss": 0.8073,
+      "step": 9331
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.425733521114396e-07,
+      "loss": 0.8481,
+      "step": 9332
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.415613577151346e-07,
+      "loss": 0.8585,
+      "step": 9333
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.405502817059937e-07,
+      "loss": 0.8664,
+      "step": 9334
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.395401241821951e-07,
+      "loss": 0.8153,
+      "step": 9335
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.385308852418191e-07,
+      "loss": 0.8088,
+      "step": 9336
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.375225649828597e-07,
+      "loss": 0.8423,
+      "step": 9337
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.365151635032218e-07,
+      "loss": 0.7789,
+      "step": 9338
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.355086809007226e-07,
+      "loss": 0.8065,
+      "step": 9339
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.345031172730875e-07,
+      "loss": 0.8836,
+      "step": 9340
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.33498472717956e-07,
+      "loss": 0.7656,
+      "step": 9341
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.324947473328735e-07,
+      "loss": 0.8136,
+      "step": 9342
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.314919412153007e-07,
+      "loss": 0.856,
+      "step": 9343
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.304900544626046e-07,
+      "loss": 0.8588,
+      "step": 9344
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.294890871720682e-07,
+      "loss": 0.781,
+      "step": 9345
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.284890394408826e-07,
+      "loss": 0.88,
+      "step": 9346
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.274899113661458e-07,
+      "loss": 0.8763,
+      "step": 9347
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.264917030448757e-07,
+      "loss": 0.8921,
+      "step": 9348
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.254944145739915e-07,
+      "loss": 0.8779,
+      "step": 9349
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.244980460503268e-07,
+      "loss": 0.8635,
+      "step": 9350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.235025975706298e-07,
+      "loss": 0.8259,
+      "step": 9351
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.225080692315532e-07,
+      "loss": 0.8781,
+      "step": 9352
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.215144611296618e-07,
+      "loss": 0.8098,
+      "step": 9353
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.205217733614353e-07,
+      "loss": 0.8978,
+      "step": 9354
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.195300060232589e-07,
+      "loss": 0.8483,
+      "step": 9355
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.185391592114286e-07,
+      "loss": 0.8704,
+      "step": 9356
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.175492330221566e-07,
+      "loss": 0.8875,
+      "step": 9357
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.165602275515592e-07,
+      "loss": 0.8864,
+      "step": 9358
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.155721428956661e-07,
+      "loss": 0.8418,
+      "step": 9359
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.145849791504187e-07,
+      "loss": 0.9063,
+      "step": 9360
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.135987364116668e-07,
+      "loss": 0.7642,
+      "step": 9361
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.126134147751716e-07,
+      "loss": 0.8493,
+      "step": 9362
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.116290143366042e-07,
+      "loss": 0.881,
+      "step": 9363
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.106455351915507e-07,
+      "loss": 0.7798,
+      "step": 9364
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.096629774355011e-07,
+      "loss": 0.8022,
+      "step": 9365
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.086813411638581e-07,
+      "loss": 0.8905,
+      "step": 9366
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.077006264719387e-07,
+      "loss": 0.8333,
+      "step": 9367
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.067208334549656e-07,
+      "loss": 0.9034,
+      "step": 9368
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.057419622080728e-07,
+      "loss": 0.8466,
+      "step": 9369
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.047640128263087e-07,
+      "loss": 0.8212,
+      "step": 9370
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.037869854046284e-07,
+      "loss": 0.9178,
+      "step": 9371
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.028108800378961e-07,
+      "loss": 0.9177,
+      "step": 9372
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.018356968208915e-07,
+      "loss": 0.863,
+      "step": 9373
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.008614358483021e-07,
+      "loss": 0.8677,
+      "step": 9374
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.998880972147246e-07,
+      "loss": 0.8299,
+      "step": 9375
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.989156810146667e-07,
+      "loss": 0.8731,
+      "step": 9376
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.979441873425495e-07,
+      "loss": 0.8604,
+      "step": 9377
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.969736162927019e-07,
+      "loss": 0.8051,
+      "step": 9378
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.960039679593609e-07,
+      "loss": 0.8322,
+      "step": 9379
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.9503524243668e-07,
+      "loss": 0.7995,
+      "step": 9380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.940674398187173e-07,
+      "loss": 0.8163,
+      "step": 9381
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.931005601994432e-07,
+      "loss": 0.8766,
+      "step": 9382
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.921346036727425e-07,
+      "loss": 0.831,
+      "step": 9383
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.911695703324038e-07,
+      "loss": 0.8916,
+      "step": 9384
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.902054602721273e-07,
+      "loss": 0.904,
+      "step": 9385
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.892422735855284e-07,
+      "loss": 0.9227,
+      "step": 9386
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.882800103661311e-07,
+      "loss": 0.8711,
+      "step": 9387
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.873186707073663e-07,
+      "loss": 0.8842,
+      "step": 9388
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.863582547025746e-07,
+      "loss": 0.7921,
+      "step": 9389
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.853987624450151e-07,
+      "loss": 0.8486,
+      "step": 9390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.844401940278487e-07,
+      "loss": 0.7792,
+      "step": 9391
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.834825495441475e-07,
+      "loss": 0.8495,
+      "step": 9392
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.825258290869017e-07,
+      "loss": 0.8568,
+      "step": 9393
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.815700327490014e-07,
+      "loss": 0.9014,
+      "step": 9394
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.806151606232523e-07,
+      "loss": 0.8213,
+      "step": 9395
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.796612128023726e-07,
+      "loss": 0.8756,
+      "step": 9396
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.787081893789835e-07,
+      "loss": 0.7899,
+      "step": 9397
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.777560904456236e-07,
+      "loss": 0.9115,
+      "step": 9398
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7680491609473966e-07,
+      "loss": 0.9193,
+      "step": 9399
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7585466641868696e-07,
+      "loss": 0.8727,
+      "step": 9400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7490534150973045e-07,
+      "loss": 0.9027,
+      "step": 9401
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7395694146004976e-07,
+      "loss": 0.8908,
+      "step": 9402
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.730094663617313e-07,
+      "loss": 0.846,
+      "step": 9403
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7206291630677024e-07,
+      "loss": 0.8259,
+      "step": 9404
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7111729138707316e-07,
+      "loss": 0.9385,
+      "step": 9405
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7017259169446104e-07,
+      "loss": 0.7585,
+      "step": 9406
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.692288173206605e-07,
+      "loss": 0.8265,
+      "step": 9407
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6828596835730487e-07,
+      "loss": 0.7526,
+      "step": 9408
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6734404489594874e-07,
+      "loss": 0.8211,
+      "step": 9409
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.664030470280467e-07,
+      "loss": 0.8138,
+      "step": 9410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6546297484496574e-07,
+      "loss": 0.9007,
+      "step": 9411
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.645238284379883e-07,
+      "loss": 0.859,
+      "step": 9412
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6358560789829923e-07,
+      "loss": 0.8668,
+      "step": 9413
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.626483133169968e-07,
+      "loss": 0.7997,
+      "step": 9414
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6171194478509264e-07,
+      "loss": 0.8605,
+      "step": 9415
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.60776502393504e-07,
+      "loss": 0.8474,
+      "step": 9416
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5984198623305833e-07,
+      "loss": 0.7724,
+      "step": 9417
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5890839639449514e-07,
+      "loss": 0.8474,
+      "step": 9418
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.579757329684653e-07,
+      "loss": 0.9054,
+      "step": 9419
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5704399604552417e-07,
+      "loss": 0.8467,
+      "step": 9420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5611318571614273e-07,
+      "loss": 0.8125,
+      "step": 9421
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.551833020707008e-07,
+      "loss": 0.8542,
+      "step": 9422
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.542543451994874e-07,
+      "loss": 0.7308,
+      "step": 9423
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.533263151926981e-07,
+      "loss": 0.8645,
+      "step": 9424
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.523992121404464e-07,
+      "loss": 0.8927,
+      "step": 9425
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.514730361327502e-07,
+      "loss": 0.8232,
+      "step": 9426
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.505477872595354e-07,
+      "loss": 0.8633,
+      "step": 9427
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4962346561064574e-07,
+      "loss": 0.8641,
+      "step": 9428
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4870007127582715e-07,
+      "loss": 0.8236,
+      "step": 9429
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4777760434473796e-07,
+      "loss": 0.8077,
+      "step": 9430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4685606490694976e-07,
+      "loss": 0.8654,
+      "step": 9431
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4593545305193774e-07,
+      "loss": 0.862,
+      "step": 9432
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4501576886909593e-07,
+      "loss": 0.8002,
+      "step": 9433
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.440970124477173e-07,
+      "loss": 0.853,
+      "step": 9434
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.43179183877015e-07,
+      "loss": 0.8242,
+      "step": 9435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4226228324610544e-07,
+      "loss": 0.8744,
+      "step": 9436
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4134631064401633e-07,
+      "loss": 0.8368,
+      "step": 9437
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.404312661596877e-07,
+      "loss": 0.8378,
+      "step": 9438
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.395171498819673e-07,
+      "loss": 0.8418,
+      "step": 9439
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.386039618996119e-07,
+      "loss": 0.7968,
+      "step": 9440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3769170230129055e-07,
+      "loss": 0.8776,
+      "step": 9441
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3678037117558135e-07,
+      "loss": 0.8008,
+      "step": 9442
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.358699686109702e-07,
+      "loss": 0.8153,
+      "step": 9443
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.349604946958563e-07,
+      "loss": 0.8033,
+      "step": 9444
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.340519495185469e-07,
+      "loss": 0.8789,
+      "step": 9445
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.331443331672591e-07,
+      "loss": 0.9216,
+      "step": 9446
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3223764573011804e-07,
+      "loss": 0.8862,
+      "step": 9447
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.313318872951633e-07,
+      "loss": 0.9066,
+      "step": 9448
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3042705795034e-07,
+      "loss": 0.8041,
+      "step": 9449
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.295231577835024e-07,
+      "loss": 0.806,
+      "step": 9450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2862018688242026e-07,
+      "loss": 0.8208,
+      "step": 9451
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2771814533476787e-07,
+      "loss": 0.831,
+      "step": 9452
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.268170332281285e-07,
+      "loss": 0.883,
+      "step": 9453
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2591685065000223e-07,
+      "loss": 0.8075,
+      "step": 9454
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2501759768779017e-07,
+      "loss": 0.8407,
+      "step": 9455
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.241192744288092e-07,
+      "loss": 0.8537,
+      "step": 9456
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2322188096028505e-07,
+      "loss": 0.789,
+      "step": 9457
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.223254173693492e-07,
+      "loss": 0.768,
+      "step": 9458
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.214298837430464e-07,
+      "loss": 0.8718,
+      "step": 9459
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2053528016833267e-07,
+      "loss": 0.8917,
+      "step": 9460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.196416067320697e-07,
+      "loss": 0.8484,
+      "step": 9461
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1874886352103015e-07,
+      "loss": 0.8154,
+      "step": 9462
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.178570506218971e-07,
+      "loss": 0.771,
+      "step": 9463
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1696616812126333e-07,
+      "loss": 0.8788,
+      "step": 9464
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.16076216105632e-07,
+      "loss": 0.8654,
+      "step": 9465
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1518719466141165e-07,
+      "loss": 0.7478,
+      "step": 9466
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1429910387492667e-07,
+      "loss": 0.8486,
+      "step": 9467
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.13411943832408e-07,
+      "loss": 0.841,
+      "step": 9468
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1252571461999457e-07,
+      "loss": 0.8492,
+      "step": 9469
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.116404163237386e-07,
+      "loss": 0.9047,
+      "step": 9470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.107560490295992e-07,
+      "loss": 0.772,
+      "step": 9471
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0987261282344425e-07,
+      "loss": 0.8507,
+      "step": 9472
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0899010779105633e-07,
+      "loss": 0.8047,
+      "step": 9473
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.081085340181223e-07,
+      "loss": 0.7751,
+      "step": 9474
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0722789159023945e-07,
+      "loss": 0.8794,
+      "step": 9475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.06348180592917e-07,
+      "loss": 0.9322,
+      "step": 9476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.054694011115723e-07,
+      "loss": 0.8431,
+      "step": 9477
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0459155323153034e-07,
+      "loss": 0.8141,
+      "step": 9478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.037146370380296e-07,
+      "loss": 0.7528,
+      "step": 9479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.028386526162176e-07,
+      "loss": 0.823,
+      "step": 9480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.019636000511473e-07,
+      "loss": 0.8164,
+      "step": 9481
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.010894794277831e-07,
+      "loss": 0.8395,
+      "step": 9482
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0021629083100254e-07,
+      "loss": 0.8013,
+      "step": 9483
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.993440343455879e-07,
+      "loss": 0.7795,
+      "step": 9484
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.984727100562313e-07,
+      "loss": 0.8409,
+      "step": 9485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.976023180475397e-07,
+      "loss": 0.8416,
+      "step": 9486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.967328584040242e-07,
+      "loss": 0.8142,
+      "step": 9487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.95864331210104e-07,
+      "loss": 0.813,
+      "step": 9488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9499673655011396e-07,
+      "loss": 0.8289,
+      "step": 9489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.941300745082932e-07,
+      "loss": 0.8928,
+      "step": 9490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.932643451687934e-07,
+      "loss": 0.8231,
+      "step": 9491
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9239954861567177e-07,
+      "loss": 0.8338,
+      "step": 9492
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9153568493290217e-07,
+      "loss": 0.8468,
+      "step": 9493
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.906727542043598e-07,
+      "loss": 0.8029,
+      "step": 9494
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8981075651383317e-07,
+      "loss": 0.8039,
+      "step": 9495
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8894969194502083e-07,
+      "loss": 0.9079,
+      "step": 9496
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.880895605815294e-07,
+      "loss": 0.8946,
+      "step": 9497
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.87230362506873e-07,
+      "loss": 0.8367,
+      "step": 9498
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8637209780448047e-07,
+      "loss": 0.8745,
+      "step": 9499
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8551476655768527e-07,
+      "loss": 0.9099,
+      "step": 9500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.846583688497307e-07,
+      "loss": 0.9153,
+      "step": 9501
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8380290476377255e-07,
+      "loss": 0.8912,
+      "step": 9502
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8294837438287435e-07,
+      "loss": 0.8572,
+      "step": 9503
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8209477779000637e-07,
+      "loss": 0.841,
+      "step": 9504
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.812421150680512e-07,
+      "loss": 0.8283,
+      "step": 9505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.803903862998004e-07,
+      "loss": 0.9002,
+      "step": 9506
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.795395915679545e-07,
+      "loss": 0.8595,
+      "step": 9507
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7868973095512185e-07,
+      "loss": 0.7752,
+      "step": 9508
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.778408045438231e-07,
+      "loss": 0.8499,
+      "step": 9509
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7699281241648565e-07,
+      "loss": 0.8292,
+      "step": 9510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.76145754655447e-07,
+      "loss": 0.826,
+      "step": 9511
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7529963134295466e-07,
+      "loss": 0.8358,
+      "step": 9512
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7445444256116403e-07,
+      "loss": 0.8278,
+      "step": 9513
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.736101883921406e-07,
+      "loss": 0.7525,
+      "step": 9514
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7276686891786093e-07,
+      "loss": 0.8317,
+      "step": 9515
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.719244842202074e-07,
+      "loss": 0.8656,
+      "step": 9516
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.710830343809713e-07,
+      "loss": 0.8498,
+      "step": 9517
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.702425194818582e-07,
+      "loss": 0.8752,
+      "step": 9518
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6940293960447737e-07,
+      "loss": 0.8439,
+      "step": 9519
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.685642948303503e-07,
+      "loss": 0.8217,
+      "step": 9520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.677265852409084e-07,
+      "loss": 0.8347,
+      "step": 9521
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.66889810917489e-07,
+      "loss": 0.8715,
+      "step": 9522
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.660539719413414e-07,
+      "loss": 0.8988,
+      "step": 9523
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6521906839362187e-07,
+      "loss": 0.8158,
+      "step": 9524
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.643851003553978e-07,
+      "loss": 0.8112,
+      "step": 9525
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.635520679076465e-07,
+      "loss": 0.8524,
+      "step": 9526
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6271997113125126e-07,
+      "loss": 0.8437,
+      "step": 9527
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6188881010700725e-07,
+      "loss": 0.9161,
+      "step": 9528
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6105858491561787e-07,
+      "loss": 0.8428,
+      "step": 9529
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6022929563769513e-07,
+      "loss": 0.8389,
+      "step": 9530
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.594009423537603e-07,
+      "loss": 0.8977,
+      "step": 9531
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5857352514424573e-07,
+      "loss": 0.8896,
+      "step": 9532
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5774704408948814e-07,
+      "loss": 0.8421,
+      "step": 9533
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5692149926974006e-07,
+      "loss": 0.883,
+      "step": 9534
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5609689076515743e-07,
+      "loss": 0.8395,
+      "step": 9535
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.552732186558072e-07,
+      "loss": 0.8449,
+      "step": 9536
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.544504830216655e-07,
+      "loss": 0.8849,
+      "step": 9537
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.536286839426195e-07,
+      "loss": 0.7555,
+      "step": 9538
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5280782149846203e-07,
+      "loss": 0.8329,
+      "step": 9539
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.51987895768896e-07,
+      "loss": 0.7839,
+      "step": 9540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5116890683353557e-07,
+      "loss": 0.7965,
+      "step": 9541
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5035085477190143e-07,
+      "loss": 0.8376,
+      "step": 9542
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.495337396634224e-07,
+      "loss": 0.8947,
+      "step": 9543
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4871756158744054e-07,
+      "loss": 0.8136,
+      "step": 9544
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.479023206232035e-07,
+      "loss": 0.8341,
+      "step": 9545
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4708801684986693e-07,
+      "loss": 0.8619,
+      "step": 9546
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4627465034649976e-07,
+      "loss": 0.821,
+      "step": 9547
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.454622211920766e-07,
+      "loss": 0.8904,
+      "step": 9548
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.446507294654833e-07,
+      "loss": 0.8793,
+      "step": 9549
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4384017524551116e-07,
+      "loss": 0.8083,
+      "step": 9550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.430305586108651e-07,
+      "loss": 0.851,
+      "step": 9551
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.422218796401544e-07,
+      "loss": 0.9385,
+      "step": 9552
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4141413841189964e-07,
+      "loss": 0.7864,
+      "step": 9553
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4060733500453247e-07,
+      "loss": 0.9099,
+      "step": 9554
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3980146949638915e-07,
+      "loss": 0.8332,
+      "step": 9555
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.38996541965716e-07,
+      "loss": 0.8466,
+      "step": 9556
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.381925524906715e-07,
+      "loss": 0.8053,
+      "step": 9557
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3738950114932e-07,
+      "loss": 0.9033,
+      "step": 9558
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.365873880196335e-07,
+      "loss": 0.8035,
+      "step": 9559
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3578621317949755e-07,
+      "loss": 0.7863,
+      "step": 9560
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3498597670670317e-07,
+      "loss": 0.8336,
+      "step": 9561
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.341866786789505e-07,
+      "loss": 0.7757,
+      "step": 9562
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3338831917384963e-07,
+      "loss": 0.8805,
+      "step": 9563
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.325908982689185e-07,
+      "loss": 0.8823,
+      "step": 9564
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3179441604158536e-07,
+      "loss": 0.8662,
+      "step": 9565
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.309988725691837e-07,
+      "loss": 0.8712,
+      "step": 9566
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.302042679289619e-07,
+      "loss": 0.8227,
+      "step": 9567
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.294106021980714e-07,
+      "loss": 0.8731,
+      "step": 9568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2861787545357514e-07,
+      "loss": 0.9022,
+      "step": 9569
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.278260877724471e-07,
+      "loss": 0.8817,
+      "step": 9570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2703523923156455e-07,
+      "loss": 0.8429,
+      "step": 9571
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2624532990771507e-07,
+      "loss": 0.8797,
+      "step": 9572
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.254563598776028e-07,
+      "loss": 0.8786,
+      "step": 9573
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2466832921782986e-07,
+      "loss": 0.8592,
+      "step": 9574
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.238812380049117e-07,
+      "loss": 0.8802,
+      "step": 9575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2309508631527486e-07,
+      "loss": 0.9189,
+      "step": 9576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2230987422525173e-07,
+      "loss": 0.789,
+      "step": 9577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.215256018110824e-07,
+      "loss": 0.8659,
+      "step": 9578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2074226914891927e-07,
+      "loss": 0.8409,
+      "step": 9579
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.199598763148215e-07,
+      "loss": 0.8654,
+      "step": 9580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.19178423384755e-07,
+      "loss": 0.824,
+      "step": 9581
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.183979104346002e-07,
+      "loss": 0.8432,
+      "step": 9582
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1761833754014096e-07,
+      "loss": 0.8168,
+      "step": 9583
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1683970477706994e-07,
+      "loss": 0.8726,
+      "step": 9584
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.160620122209912e-07,
+      "loss": 0.8888,
+      "step": 9585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1528525994741876e-07,
+      "loss": 0.8817,
+      "step": 9586
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.145094480317712e-07,
+      "loss": 0.87,
+      "step": 9587
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.13734576549376e-07,
+      "loss": 0.8688,
+      "step": 9588
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.129606455754741e-07,
+      "loss": 0.8597,
+      "step": 9589
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.121876551852099e-07,
+      "loss": 0.8966,
+      "step": 9590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1141560545363883e-07,
+      "loss": 0.9056,
+      "step": 9591
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1064449645572536e-07,
+      "loss": 0.8965,
+      "step": 9592
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0987432826634076e-07,
+      "loss": 0.7902,
+      "step": 9593
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.091051009602675e-07,
+      "loss": 0.9031,
+      "step": 9594
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0833681461219344e-07,
+      "loss": 0.8381,
+      "step": 9595
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0756946929672017e-07,
+      "loss": 0.8846,
+      "step": 9596
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0680306508835244e-07,
+      "loss": 0.8555,
+      "step": 9597
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.060376020615052e-07,
+      "loss": 0.9241,
+      "step": 9598
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.052730802905046e-07,
+      "loss": 0.8174,
+      "step": 9599
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0450949984958347e-07,
+      "loss": 0.8853,
+      "step": 9600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.037468608128802e-07,
+      "loss": 0.8595,
+      "step": 9601
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0298516325444893e-07,
+      "loss": 0.842,
+      "step": 9602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.022244072482461e-07,
+      "loss": 0.8557,
+      "step": 9603
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0146459286813924e-07,
+      "loss": 0.804,
+      "step": 9604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0070572018790376e-07,
+      "loss": 0.7993,
+      "step": 9605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.999477892812264e-07,
+      "loss": 0.817,
+      "step": 9606
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.991908002216959e-07,
+      "loss": 0.9505,
+      "step": 9607
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.984347530828158e-07,
+      "loss": 0.8583,
+      "step": 9608
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.976796479379973e-07,
+      "loss": 0.8918,
+      "step": 9609
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.969254848605585e-07,
+      "loss": 0.8406,
+      "step": 9610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.961722639237241e-07,
+      "loss": 0.8535,
+      "step": 9611
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9541998520063344e-07,
+      "loss": 0.8739,
+      "step": 9612
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.94668648764328e-07,
+      "loss": 0.8258,
+      "step": 9613
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9391825468775946e-07,
+      "loss": 0.7638,
+      "step": 9614
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.931688030437929e-07,
+      "loss": 0.8896,
+      "step": 9615
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9242029390519454e-07,
+      "loss": 0.8626,
+      "step": 9616
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.916727273446418e-07,
+      "loss": 0.7889,
+      "step": 9617
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.909261034347255e-07,
+      "loss": 0.8681,
+      "step": 9618
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.901804222479354e-07,
+      "loss": 0.9083,
+      "step": 9619
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.894356838566792e-07,
+      "loss": 0.8872,
+      "step": 9620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.886918883332668e-07,
+      "loss": 0.9199,
+      "step": 9621
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.879490357499204e-07,
+      "loss": 0.7965,
+      "step": 9622
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8720712617876677e-07,
+      "loss": 0.8593,
+      "step": 9623
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.864661596918428e-07,
+      "loss": 0.8806,
+      "step": 9624
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8572613636109636e-07,
+      "loss": 0.8765,
+      "step": 9625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.849870562583812e-07,
+      "loss": 0.8947,
+      "step": 9626
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8424891945545763e-07,
+      "loss": 0.8644,
+      "step": 9627
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8351172602399945e-07,
+      "loss": 0.8472,
+      "step": 9628
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8277547603558496e-07,
+      "loss": 0.9346,
+      "step": 9629
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8204016956169924e-07,
+      "loss": 0.9143,
+      "step": 9630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.81305806673744e-07,
+      "loss": 0.8793,
+      "step": 9631
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8057238744301994e-07,
+      "loss": 0.7974,
+      "step": 9632
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7983991194074025e-07,
+      "loss": 0.9084,
+      "step": 9633
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7910838023802676e-07,
+      "loss": 0.8183,
+      "step": 9634
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7837779240590943e-07,
+      "loss": 0.9305,
+      "step": 9635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7764814851532485e-07,
+      "loss": 0.8674,
+      "step": 9636
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7691944863712185e-07,
+      "loss": 0.888,
+      "step": 9637
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.761916928420527e-07,
+      "loss": 0.8717,
+      "step": 9638
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.75464881200781e-07,
+      "loss": 0.8329,
+      "step": 9639
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.74739013783879e-07,
+      "loss": 0.7807,
+      "step": 9640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7401409066182603e-07,
+      "loss": 0.8452,
+      "step": 9641
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7329011190500797e-07,
+      "loss": 0.8717,
+      "step": 9642
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.72567077583723e-07,
+      "loss": 0.8045,
+      "step": 9643
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7184498776817615e-07,
+      "loss": 0.9055,
+      "step": 9644
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7112384252848013e-07,
+      "loss": 0.8658,
+      "step": 9645
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.704036419346534e-07,
+      "loss": 0.7904,
+      "step": 9646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.696843860566301e-07,
+      "loss": 0.8776,
+      "step": 9647
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.689660749642442e-07,
+      "loss": 0.9293,
+      "step": 9648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6824870872724116e-07,
+      "loss": 0.86,
+      "step": 9649
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.675322874152786e-07,
+      "loss": 0.891,
+      "step": 9650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6681681109791746e-07,
+      "loss": 0.8466,
+      "step": 9651
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6610227984462556e-07,
+      "loss": 0.7768,
+      "step": 9652
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.653886937247863e-07,
+      "loss": 0.7895,
+      "step": 9653
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.646760528076842e-07,
+      "loss": 0.7948,
+      "step": 9654
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.63964357162515e-07,
+      "loss": 0.8038,
+      "step": 9655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6325360685838243e-07,
+      "loss": 0.9078,
+      "step": 9656
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.625438019643001e-07,
+      "loss": 0.8929,
+      "step": 9657
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.618349425491851e-07,
+      "loss": 0.8556,
+      "step": 9658
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.611270286818668e-07,
+      "loss": 0.9097,
+      "step": 9659
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.604200604310825e-07,
+      "loss": 0.8848,
+      "step": 9660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.597140378654761e-07,
+      "loss": 0.8215,
+      "step": 9661
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.590089610535984e-07,
+      "loss": 0.8536,
+      "step": 9662
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.583048300639146e-07,
+      "loss": 0.829,
+      "step": 9663
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5760164496479e-07,
+      "loss": 0.7933,
+      "step": 9664
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5689940582450334e-07,
+      "loss": 0.8441,
+      "step": 9665
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5619811271123897e-07,
+      "loss": 0.8125,
+      "step": 9666
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5549776569309347e-07,
+      "loss": 0.9026,
+      "step": 9667
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5479836483806586e-07,
+      "loss": 0.9068,
+      "step": 9668
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.540999102140651e-07,
+      "loss": 0.9409,
+      "step": 9669
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5340240188891143e-07,
+      "loss": 0.8857,
+      "step": 9670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.527058399303306e-07,
+      "loss": 0.8511,
+      "step": 9671
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.520102244059552e-07,
+      "loss": 0.8386,
+      "step": 9672
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.513155553833291e-07,
+      "loss": 0.7939,
+      "step": 9673
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.506218329299026e-07,
+      "loss": 0.8882,
+      "step": 9674
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.499290571130319e-07,
+      "loss": 0.8244,
+      "step": 9675
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4923722799998664e-07,
+      "loss": 0.8545,
+      "step": 9676
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.485463456579396e-07,
+      "loss": 0.8593,
+      "step": 9677
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4785641015397375e-07,
+      "loss": 0.8376,
+      "step": 9678
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.47167421555079e-07,
+      "loss": 0.8886,
+      "step": 9679
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.464793799281573e-07,
+      "loss": 0.8826,
+      "step": 9680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4579228534001185e-07,
+      "loss": 0.8328,
+      "step": 9681
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4510613785735936e-07,
+      "loss": 0.8037,
+      "step": 9682
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4442093754682316e-07,
+      "loss": 0.8031,
+      "step": 9683
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4373668447493225e-07,
+      "loss": 0.7414,
+      "step": 9684
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4305337870812793e-07,
+      "loss": 0.9173,
+      "step": 9685
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.423710203127561e-07,
+      "loss": 0.8887,
+      "step": 9686
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4168960935507134e-07,
+      "loss": 0.8765,
+      "step": 9687
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.410091459012376e-07,
+      "loss": 0.8399,
+      "step": 9688
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4032963001732523e-07,
+      "loss": 0.7977,
+      "step": 9689
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3965106176931375e-07,
+      "loss": 0.8333,
+      "step": 9690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.389734412230904e-07,
+      "loss": 0.8524,
+      "step": 9691
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3829676844444926e-07,
+      "loss": 0.8547,
+      "step": 9692
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3762104349909553e-07,
+      "loss": 0.893,
+      "step": 9693
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3694626645263675e-07,
+      "loss": 0.8439,
+      "step": 9694
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.362724373705938e-07,
+      "loss": 0.8399,
+      "step": 9695
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3559955631839436e-07,
+      "loss": 0.8543,
+      "step": 9696
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3492762336137065e-07,
+      "loss": 0.7989,
+      "step": 9697
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3425663856476932e-07,
+      "loss": 0.8264,
+      "step": 9698
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3358660199373717e-07,
+      "loss": 0.8036,
+      "step": 9699
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3291751371333438e-07,
+      "loss": 0.813,
+      "step": 9700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3224937378852675e-07,
+      "loss": 0.8419,
+      "step": 9701
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3158218228419127e-07,
+      "loss": 0.9364,
+      "step": 9702
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3091593926510725e-07,
+      "loss": 0.8705,
+      "step": 9703
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3025064479596625e-07,
+      "loss": 0.8707,
+      "step": 9704
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2958629894136775e-07,
+      "loss": 0.8443,
+      "step": 9705
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2892290176581678e-07,
+      "loss": 0.824,
+      "step": 9706
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2826045333372514e-07,
+      "loss": 0.8558,
+      "step": 9707
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2759895370941809e-07,
+      "loss": 0.848,
+      "step": 9708
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.269384029571231e-07,
+      "loss": 0.8431,
+      "step": 9709
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2627880114097779e-07,
+      "loss": 0.8101,
+      "step": 9710
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2562014832502977e-07,
+      "loss": 0.8054,
+      "step": 9711
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2496244457323013e-07,
+      "loss": 0.8486,
+      "step": 9712
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2430568994944112e-07,
+      "loss": 0.7678,
+      "step": 9713
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2364988451742953e-07,
+      "loss": 0.9119,
+      "step": 9714
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2299502834087662e-07,
+      "loss": 0.8384,
+      "step": 9715
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2234112148336373e-07,
+      "loss": 0.9325,
+      "step": 9716
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2168816400838232e-07,
+      "loss": 0.7993,
+      "step": 9717
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2103615597933613e-07,
+      "loss": 0.878,
+      "step": 9718
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2038509745953118e-07,
+      "loss": 0.8463,
+      "step": 9719
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1973498851218244e-07,
+      "loss": 0.7562,
+      "step": 9720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.190858292004161e-07,
+      "loss": 0.8864,
+      "step": 9721
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1843761958726283e-07,
+      "loss": 0.8139,
+      "step": 9722
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.177903597356601e-07,
+      "loss": 0.7927,
+      "step": 9723
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1714404970845647e-07,
+      "loss": 0.878,
+      "step": 9724
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1649868956840847e-07,
+      "loss": 0.8823,
+      "step": 9725
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1585427937817594e-07,
+      "loss": 0.7729,
+      "step": 9726
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1521081920032992e-07,
+      "loss": 0.78,
+      "step": 9727
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.145683090973494e-07,
+      "loss": 0.8605,
+      "step": 9728
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1392674913161992e-07,
+      "loss": 0.784,
+      "step": 9729
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1328613936543396e-07,
+      "loss": 0.8294,
+      "step": 9730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1264647986099507e-07,
+      "loss": 0.8674,
+      "step": 9731
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1200777068041134e-07,
+      "loss": 0.8916,
+      "step": 9732
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1137001188569763e-07,
+      "loss": 0.8845,
+      "step": 9733
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1073320353878102e-07,
+      "loss": 0.865,
+      "step": 9734
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1009734570149322e-07,
+      "loss": 0.8802,
+      "step": 9735
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0946243843557367e-07,
+      "loss": 0.8171,
+      "step": 9736
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0882848180267091e-07,
+      "loss": 0.7814,
+      "step": 9737
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0819547586434008e-07,
+      "loss": 0.9381,
+      "step": 9738
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0756342068204316e-07,
+      "loss": 0.8316,
+      "step": 9739
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0693231631715105e-07,
+      "loss": 0.8573,
+      "step": 9740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0630216283094363e-07,
+      "loss": 0.8681,
+      "step": 9741
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0567296028460638e-07,
+      "loss": 0.8277,
+      "step": 9742
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0504470873923153e-07,
+      "loss": 0.7934,
+      "step": 9743
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0441740825582258e-07,
+      "loss": 0.8818,
+      "step": 9744
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.037910588952885e-07,
+      "loss": 0.8277,
+      "step": 9745
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0316566071844402e-07,
+      "loss": 0.8541,
+      "step": 9746
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0254121378601388e-07,
+      "loss": 0.8998,
+      "step": 9747
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0191771815863292e-07,
+      "loss": 0.9458,
+      "step": 9748
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0129517389683718e-07,
+      "loss": 0.8524,
+      "step": 9749
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0067358106107714e-07,
+      "loss": 0.8228,
+      "step": 9750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0005293971170568e-07,
+      "loss": 0.7998,
+      "step": 9751
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.994332499089846e-07,
+      "loss": 0.7992,
+      "step": 9752
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9881451171308686e-07,
+      "loss": 0.8212,
+      "step": 9753
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9819672518408662e-07,
+      "loss": 0.8096,
+      "step": 9754
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9757989038197146e-07,
+      "loss": 0.8727,
+      "step": 9755
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9696400736663457e-07,
+      "loss": 0.8078,
+      "step": 9756
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.963490761978748e-07,
+      "loss": 0.8708,
+      "step": 9757
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9573509693540104e-07,
+      "loss": 0.8095,
+      "step": 9758
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9512206963882786e-07,
+      "loss": 0.807,
+      "step": 9759
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9450999436768093e-07,
+      "loss": 0.8266,
+      "step": 9760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.938988711813894e-07,
+      "loss": 0.8269,
+      "step": 9761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9328870013929134e-07,
+      "loss": 0.8098,
+      "step": 9762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9267948130063274e-07,
+      "loss": 0.8732,
+      "step": 9763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9207121472456846e-07,
+      "loss": 0.8761,
+      "step": 9764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9146390047015573e-07,
+      "loss": 0.8669,
+      "step": 9765
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9085753859636736e-07,
+      "loss": 0.7363,
+      "step": 9766
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9025212916207737e-07,
+      "loss": 0.9172,
+      "step": 9767
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8964767222606873e-07,
+      "loss": 0.8539,
+      "step": 9768
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8904416784703338e-07,
+      "loss": 0.881,
+      "step": 9769
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8844161608356782e-07,
+      "loss": 0.8402,
+      "step": 9770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8784001699418186e-07,
+      "loss": 0.8058,
+      "step": 9771
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.872393706372866e-07,
+      "loss": 0.778,
+      "step": 9772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8663967707120313e-07,
+      "loss": 0.8441,
+      "step": 9773
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8604093635416155e-07,
+      "loss": 0.8608,
+      "step": 9774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8544314854429425e-07,
+      "loss": 0.9117,
+      "step": 9775
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8484631369964922e-07,
+      "loss": 0.7966,
+      "step": 9776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.842504318781746e-07,
+      "loss": 0.8806,
+      "step": 9777
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8365550313772852e-07,
+      "loss": 0.7441,
+      "step": 9778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.830615275360792e-07,
+      "loss": 0.8467,
+      "step": 9779
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8246850513089832e-07,
+      "loss": 0.8464,
+      "step": 9780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8187643597976645e-07,
+      "loss": 0.8229,
+      "step": 9781
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8128532014017098e-07,
+      "loss": 0.858,
+      "step": 9782
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8069515766951152e-07,
+      "loss": 0.8903,
+      "step": 9783
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8010594862508669e-07,
+      "loss": 0.8213,
+      "step": 9784
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.795176930641085e-07,
+      "loss": 0.8202,
+      "step": 9785
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.789303910436968e-07,
+      "loss": 0.8392,
+      "step": 9786
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.783440426208738e-07,
+      "loss": 0.8786,
+      "step": 9787
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.777586478525739e-07,
+      "loss": 0.864,
+      "step": 9788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7717420679563724e-07,
+      "loss": 0.8182,
+      "step": 9789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7659071950681172e-07,
+      "loss": 0.7553,
+      "step": 9790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7600818604274984e-07,
+      "loss": 0.8121,
+      "step": 9791
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.754266064600174e-07,
+      "loss": 0.8188,
+      "step": 9792
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7484598081508264e-07,
+      "loss": 0.8124,
+      "step": 9793
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.742663091643204e-07,
+      "loss": 0.8517,
+      "step": 9794
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7368759156401905e-07,
+      "loss": 0.8272,
+      "step": 9795
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7310982807036915e-07,
+      "loss": 0.7528,
+      "step": 9796
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7253301873946803e-07,
+      "loss": 0.8168,
+      "step": 9797
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.719571636273243e-07,
+      "loss": 0.8264,
+      "step": 9798
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7138226278985093e-07,
+      "loss": 0.8089,
+      "step": 9799
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7080831628286886e-07,
+      "loss": 0.763,
+      "step": 9800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7023532416210686e-07,
+      "loss": 0.7808,
+      "step": 9801
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6966328648320152e-07,
+      "loss": 0.83,
+      "step": 9802
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6909220330169507e-07,
+      "loss": 0.8539,
+      "step": 9803
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.685220746730387e-07,
+      "loss": 0.8793,
+      "step": 9804
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6795290065258928e-07,
+      "loss": 0.8368,
+      "step": 9805
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.673846812956137e-07,
+      "loss": 0.79,
+      "step": 9806
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6681741665728225e-07,
+      "loss": 0.845,
+      "step": 9807
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6625110679267642e-07,
+      "loss": 0.8816,
+      "step": 9808
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6568575175678338e-07,
+      "loss": 0.8695,
+      "step": 9809
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6512135160449583e-07,
+      "loss": 0.8912,
+      "step": 9810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6455790639061775e-07,
+      "loss": 0.8055,
+      "step": 9811
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6399541616985648e-07,
+      "loss": 0.9221,
+      "step": 9812
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6343388099682722e-07,
+      "loss": 0.8998,
+      "step": 9813
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6287330092605525e-07,
+      "loss": 0.8311,
+      "step": 9814
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6231367601197146e-07,
+      "loss": 0.9031,
+      "step": 9815
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6175500630891128e-07,
+      "loss": 0.9029,
+      "step": 9816
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6119729187112244e-07,
+      "loss": 0.9108,
+      "step": 9817
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6064053275275716e-07,
+      "loss": 0.8571,
+      "step": 9818
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.600847290078744e-07,
+      "loss": 0.8807,
+      "step": 9819
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5952988069044105e-07,
+      "loss": 0.8916,
+      "step": 9820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5897598785433176e-07,
+      "loss": 0.8331,
+      "step": 9821
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5842305055332796e-07,
+      "loss": 0.8903,
+      "step": 9822
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.578710688411167e-07,
+      "loss": 0.8439,
+      "step": 9823
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.573200427712973e-07,
+      "loss": 0.8227,
+      "step": 9824
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5676997239736925e-07,
+      "loss": 0.8857,
+      "step": 9825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.562208577727442e-07,
+      "loss": 0.7531,
+      "step": 9826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5567269895073954e-07,
+      "loss": 0.825,
+      "step": 9827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.551254959845805e-07,
+      "loss": 0.8386,
+      "step": 9828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5457924892739784e-07,
+      "loss": 0.8914,
+      "step": 9829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.540339578322314e-07,
+      "loss": 0.7695,
+      "step": 9830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5348962275202773e-07,
+      "loss": 0.8139,
+      "step": 9831
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5294624373963894e-07,
+      "loss": 0.855,
+      "step": 9832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5240382084782622e-07,
+      "loss": 0.8537,
+      "step": 9833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5186235412925744e-07,
+      "loss": 0.8694,
+      "step": 9834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5132184363650714e-07,
+      "loss": 0.7999,
+      "step": 9835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5078228942205674e-07,
+      "loss": 0.8337,
+      "step": 9836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5024369153829767e-07,
+      "loss": 0.8219,
+      "step": 9837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4970605003752359e-07,
+      "loss": 0.8448,
+      "step": 9838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4916936497193836e-07,
+      "loss": 0.7681,
+      "step": 9839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4863363639365357e-07,
+      "loss": 0.8782,
+      "step": 9840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4809886435468768e-07,
+      "loss": 0.801,
+      "step": 9841
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4756504890696466e-07,
+      "loss": 0.8062,
+      "step": 9842
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.470321901023153e-07,
+      "loss": 0.8809,
+      "step": 9843
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4650028799247928e-07,
+      "loss": 0.7957,
+      "step": 9844
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4596934262910423e-07,
+      "loss": 0.8607,
+      "step": 9845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.454393540637411e-07,
+      "loss": 0.8181,
+      "step": 9846
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4491032234785207e-07,
+      "loss": 0.8092,
+      "step": 9847
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4438224753280384e-07,
+      "loss": 0.8419,
+      "step": 9848
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4385512966986982e-07,
+      "loss": 0.7904,
+      "step": 9849
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4332896881023462e-07,
+      "loss": 0.9242,
+      "step": 9850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4280376500498517e-07,
+      "loss": 0.8427,
+      "step": 9851
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.422795183051151e-07,
+      "loss": 0.8443,
+      "step": 9852
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.417562287615315e-07,
+      "loss": 0.8381,
+      "step": 9853
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4123389642504148e-07,
+      "loss": 0.8699,
+      "step": 9854
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4071252134636226e-07,
+      "loss": 0.8703,
+      "step": 9855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.401921035761189e-07,
+      "loss": 0.8484,
+      "step": 9856
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3967264316484207e-07,
+      "loss": 0.798,
+      "step": 9857
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3915414016296925e-07,
+      "loss": 0.9155,
+      "step": 9858
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3863659462084678e-07,
+      "loss": 0.7982,
+      "step": 9859
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.381200065887256e-07,
+      "loss": 0.9017,
+      "step": 9860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.376043761167656e-07,
+      "loss": 0.8638,
+      "step": 9861
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3708970325503222e-07,
+      "loss": 0.8631,
+      "step": 9862
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3657598805350003e-07,
+      "loss": 0.8055,
+      "step": 9863
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3606323056204795e-07,
+      "loss": 0.8331,
+      "step": 9864
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3555143083046397e-07,
+      "loss": 0.8853,
+      "step": 9865
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3504058890844274e-07,
+      "loss": 0.8282,
+      "step": 9866
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3453070484558461e-07,
+      "loss": 0.8993,
+      "step": 9867
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3402177869139887e-07,
+      "loss": 0.8672,
+      "step": 9868
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3351381049530044e-07,
+      "loss": 0.8257,
+      "step": 9869
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3300680030661096e-07,
+      "loss": 0.8009,
+      "step": 9870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3250074817455993e-07,
+      "loss": 0.8378,
+      "step": 9871
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3199565414828363e-07,
+      "loss": 0.8625,
+      "step": 9872
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3149151827682617e-07,
+      "loss": 0.8274,
+      "step": 9873
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3098834060913612e-07,
+      "loss": 0.9084,
+      "step": 9874
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3048612119407111e-07,
+      "loss": 0.8747,
+      "step": 9875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2998486008039545e-07,
+      "loss": 0.8688,
+      "step": 9876
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2948455731678133e-07,
+      "loss": 0.8244,
+      "step": 9877
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.289852129518032e-07,
+      "loss": 0.8994,
+      "step": 9878
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.284868270339501e-07,
+      "loss": 0.76,
+      "step": 9879
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2798939961161217e-07,
+      "loss": 0.9192,
+      "step": 9880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2749293073308632e-07,
+      "loss": 0.8328,
+      "step": 9881
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.269974204465818e-07,
+      "loss": 0.8986,
+      "step": 9882
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2650286880020902e-07,
+      "loss": 0.8345,
+      "step": 9883
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2600927584198618e-07,
+      "loss": 0.8692,
+      "step": 9884
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2551664161984388e-07,
+      "loss": 0.8088,
+      "step": 9885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2502496618161165e-07,
+      "loss": 0.9021,
+      "step": 9886
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2453424957503124e-07,
+      "loss": 0.8753,
+      "step": 9887
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2404449184774904e-07,
+      "loss": 0.8237,
+      "step": 9888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2355569304732029e-07,
+      "loss": 0.8486,
+      "step": 9889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2306785322120596e-07,
+      "loss": 0.8506,
+      "step": 9890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.225809724167737e-07,
+      "loss": 0.8417,
+      "step": 9891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.220950506812968e-07,
+      "loss": 0.8914,
+      "step": 9892
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2161008806195974e-07,
+      "loss": 0.9293,
+      "step": 9893
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2112608460584707e-07,
+      "loss": 0.8316,
+      "step": 9894
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.206430403599579e-07,
+      "loss": 0.8999,
+      "step": 9895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2016095537119242e-07,
+      "loss": 0.852,
+      "step": 9896
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1967982968635994e-07,
+      "loss": 0.8195,
+      "step": 9897
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1919966335217636e-07,
+      "loss": 0.8287,
+      "step": 9898
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1872045641526442e-07,
+      "loss": 0.9025,
+      "step": 9899
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1824220892215465e-07,
+      "loss": 0.8326,
+      "step": 9900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1776492091928215e-07,
+      "loss": 0.875,
+      "step": 9901
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1728859245299207e-07,
+      "loss": 0.9003,
+      "step": 9902
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1681322356953295e-07,
+      "loss": 0.8377,
+      "step": 9903
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1633881431506122e-07,
+      "loss": 0.8616,
+      "step": 9904
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1586536473564225e-07,
+      "loss": 0.8637,
+      "step": 9905
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1539287487724594e-07,
+      "loss": 0.8532,
+      "step": 9906
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.149213447857489e-07,
+      "loss": 0.8217,
+      "step": 9907
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1445077450693786e-07,
+      "loss": 0.8533,
+      "step": 9908
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1398116408650073e-07,
+      "loss": 0.908,
+      "step": 9909
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1351251357003656e-07,
+      "loss": 0.8669,
+      "step": 9910
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1304482300305008e-07,
+      "loss": 0.8989,
+      "step": 9911
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1257809243095385e-07,
+      "loss": 0.8521,
+      "step": 9912
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1211232189906384e-07,
+      "loss": 0.8376,
+      "step": 9913
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1164751145260722e-07,
+      "loss": 0.9,
+      "step": 9914
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1118366113671453e-07,
+      "loss": 0.8481,
+      "step": 9915
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.107207709964242e-07,
+      "loss": 0.9144,
+      "step": 9916
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1025884107668139e-07,
+      "loss": 0.8868,
+      "step": 9917
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0979787142233911e-07,
+      "loss": 0.751,
+      "step": 9918
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.093378620781549e-07,
+      "loss": 0.8748,
+      "step": 9919
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0887881308879633e-07,
+      "loss": 0.8601,
+      "step": 9920
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0842072449883334e-07,
+      "loss": 0.841,
+      "step": 9921
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0796359635274701e-07,
+      "loss": 0.8641,
+      "step": 9922
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0750742869492071e-07,
+      "loss": 0.7943,
+      "step": 9923
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0705222156965011e-07,
+      "loss": 0.8458,
+      "step": 9924
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0659797502113322e-07,
+      "loss": 0.9296,
+      "step": 9925
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0614468909347476e-07,
+      "loss": 0.8628,
+      "step": 9926
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0569236383068948e-07,
+      "loss": 0.8099,
+      "step": 9927
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0524099927669563e-07,
+      "loss": 0.8777,
+      "step": 9928
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0479059547531922e-07,
+      "loss": 0.7812,
+      "step": 9929
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0434115247029419e-07,
+      "loss": 0.7908,
+      "step": 9930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0389267030526007e-07,
+      "loss": 0.8208,
+      "step": 9931
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0344514902376201e-07,
+      "loss": 0.8223,
+      "step": 9932
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0299858866925416e-07,
+      "loss": 0.8071,
+      "step": 9933
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0255298928509627e-07,
+      "loss": 0.9181,
+      "step": 9934
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0210835091455373e-07,
+      "loss": 0.9182,
+      "step": 9935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0166467360079979e-07,
+      "loss": 0.8405,
+      "step": 9936
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0122195738691665e-07,
+      "loss": 0.8836,
+      "step": 9937
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.007802023158877e-07,
+      "loss": 0.8016,
+      "step": 9938
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0033940843060641e-07,
+      "loss": 0.887,
+      "step": 9939
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.989957577387521e-08,
+      "loss": 0.7845,
+      "step": 9940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.946070438839772e-08,
+      "loss": 0.8512,
+      "step": 9941
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.902279431678874e-08,
+      "loss": 0.8139,
+      "step": 9942
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.858584560156759e-08,
+      "loss": 0.8842,
+      "step": 9943
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.814985828516033e-08,
+      "loss": 0.8672,
+      "step": 9944
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.771483240989977e-08,
+      "loss": 0.7949,
+      "step": 9945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.728076801802656e-08,
+      "loss": 0.8561,
+      "step": 9946
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.684766515168809e-08,
+      "loss": 0.8274,
+      "step": 9947
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.641552385293518e-08,
+      "loss": 0.8224,
+      "step": 9948
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.59843441637287e-08,
+      "loss": 0.8826,
+      "step": 9949
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.555412612593518e-08,
+      "loss": 0.874,
+      "step": 9950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.512486978132674e-08,
+      "loss": 0.8464,
+      "step": 9951
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.469657517158226e-08,
+      "loss": 0.8594,
+      "step": 9952
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.426924233829072e-08,
+      "loss": 0.9061,
+      "step": 9953
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.384287132294223e-08,
+      "loss": 0.8928,
+      "step": 9954
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.341746216693703e-08,
+      "loss": 0.8563,
+      "step": 9955
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.299301491158207e-08,
+      "loss": 0.8604,
+      "step": 9956
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.256952959808774e-08,
+      "loss": 0.8708,
+      "step": 9957
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.214700626757667e-08,
+      "loss": 0.8164,
+      "step": 9958
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.17254449610705e-08,
+      "loss": 0.8246,
+      "step": 9959
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.130484571950538e-08,
+      "loss": 0.8297,
+      "step": 9960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.088520858371752e-08,
+      "loss": 0.8078,
+      "step": 9961
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.046653359445323e-08,
+      "loss": 0.875,
+      "step": 9962
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.004882079236666e-08,
+      "loss": 0.8777,
+      "step": 9963
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.963207021801423e-08,
+      "loss": 0.8522,
+      "step": 9964
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.921628191186027e-08,
+      "loss": 0.8374,
+      "step": 9965
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.880145591428024e-08,
+      "loss": 0.8435,
+      "step": 9966
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.838759226554971e-08,
+      "loss": 0.8179,
+      "step": 9967
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.797469100585432e-08,
+      "loss": 0.8525,
+      "step": 9968
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.756275217528532e-08,
+      "loss": 0.8995,
+      "step": 9969
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.715177581384182e-08,
+      "loss": 0.7704,
+      "step": 9970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.674176196142748e-08,
+      "loss": 0.9182,
+      "step": 9971
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.633271065785486e-08,
+      "loss": 0.88,
+      "step": 9972
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.592462194284001e-08,
+      "loss": 0.8747,
+      "step": 9973
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.551749585600678e-08,
+      "loss": 0.8022,
+      "step": 9974
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.511133243688796e-08,
+      "loss": 0.8054,
+      "step": 9975
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.470613172491981e-08,
+      "loss": 0.8183,
+      "step": 9976
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.430189375944531e-08,
+      "loss": 0.8074,
+      "step": 9977
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.389861857971748e-08,
+      "loss": 0.8461,
+      "step": 9978
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.349630622488946e-08,
+      "loss": 0.856,
+      "step": 9979
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.309495673402778e-08,
+      "loss": 0.8409,
+      "step": 9980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.269457014610016e-08,
+      "loss": 0.8366,
+      "step": 9981
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.229514649998438e-08,
+      "loss": 0.9209,
+      "step": 9982
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.189668583446276e-08,
+      "loss": 0.8664,
+      "step": 9983
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.149918818822433e-08,
+      "loss": 0.8864,
+      "step": 9984
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.110265359986603e-08,
+      "loss": 0.7885,
+      "step": 9985
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.070708210788925e-08,
+      "loss": 0.7934,
+      "step": 9986
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.031247375070328e-08,
+      "loss": 0.8221,
+      "step": 9987
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.991882856662303e-08,
+      "loss": 0.8229,
+      "step": 9988
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.952614659387126e-08,
+      "loss": 0.8477,
+      "step": 9989
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.913442787057523e-08,
+      "loss": 0.8323,
+      "step": 9990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.874367243477011e-08,
+      "loss": 0.8314,
+      "step": 9991
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.835388032439661e-08,
+      "loss": 0.8969,
+      "step": 9992
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.796505157730339e-08,
+      "loss": 0.8685,
+      "step": 9993
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.757718623124466e-08,
+      "loss": 0.8704,
+      "step": 9994
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.719028432388032e-08,
+      "loss": 0.8057,
+      "step": 9995
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.680434589277696e-08,
+      "loss": 0.8095,
+      "step": 9996
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.641937097540909e-08,
+      "loss": 0.7922,
+      "step": 9997
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.603535960915675e-08,
+      "loss": 0.8827,
+      "step": 9998
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.565231183130573e-08,
+      "loss": 0.839,
+      "step": 9999
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.527022767904957e-08,
+      "loss": 0.8497,
+      "step": 10000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.488910718948639e-08,
+      "loss": 0.8688,
+      "step": 10001
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.450895039962214e-08,
+      "loss": 0.7924,
+      "step": 10002
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.412975734636951e-08,
+      "loss": 0.8914,
+      "step": 10003
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.375152806654685e-08,
+      "loss": 0.8656,
+      "step": 10004
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.33742625968803e-08,
+      "loss": 0.8374,
+      "step": 10005
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.299796097399947e-08,
+      "loss": 0.8932,
+      "step": 10006
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.26226232344418e-08,
+      "loss": 0.8862,
+      "step": 10007
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.224824941465369e-08,
+      "loss": 0.8427,
+      "step": 10008
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.187483955098495e-08,
+      "loss": 0.8718,
+      "step": 10009
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.150239367969102e-08,
+      "loss": 0.807,
+      "step": 10010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.113091183693744e-08,
+      "loss": 0.8946,
+      "step": 10011
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.076039405879309e-08,
+      "loss": 0.9252,
+      "step": 10012
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.039084038123367e-08,
+      "loss": 0.9275,
+      "step": 10013
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.002225084014269e-08,
+      "loss": 0.7772,
+      "step": 10014
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.96546254713093e-08,
+      "loss": 0.8296,
+      "step": 10015
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.928796431042717e-08,
+      "loss": 0.881,
+      "step": 10016
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.892226739310004e-08,
+      "loss": 0.7757,
+      "step": 10017
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.855753475483507e-08,
+      "loss": 0.8535,
+      "step": 10018
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.819376643104725e-08,
+      "loss": 0.8998,
+      "step": 10019
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.783096245705612e-08,
+      "loss": 0.9272,
+      "step": 10020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.746912286809016e-08,
+      "loss": 0.7769,
+      "step": 10021
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.710824769928349e-08,
+      "loss": 0.8241,
+      "step": 10022
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.674833698567473e-08,
+      "loss": 0.891,
+      "step": 10023
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.63893907622104e-08,
+      "loss": 0.8513,
+      "step": 10024
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.603140906374372e-08,
+      "loss": 0.9283,
+      "step": 10025
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.567439192503244e-08,
+      "loss": 0.7574,
+      "step": 10026
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.531833938074328e-08,
+      "loss": 0.8384,
+      "step": 10027
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.496325146544746e-08,
+      "loss": 0.845,
+      "step": 10028
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.460912821362297e-08,
+      "loss": 0.7756,
+      "step": 10029
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.425596965965453e-08,
+      "loss": 0.9052,
+      "step": 10030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.390377583783358e-08,
+      "loss": 0.8848,
+      "step": 10031
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.35525467823539e-08,
+      "loss": 0.8546,
+      "step": 10032
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.320228252732263e-08,
+      "loss": 0.8458,
+      "step": 10033
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.285298310674703e-08,
+      "loss": 0.8321,
+      "step": 10034
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.25046485545433e-08,
+      "loss": 0.7827,
+      "step": 10035
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.215727890453438e-08,
+      "loss": 0.8745,
+      "step": 10036
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.181087419044885e-08,
+      "loss": 0.8538,
+      "step": 10037
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.14654344459209e-08,
+      "loss": 0.8299,
+      "step": 10038
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.112095970449261e-08,
+      "loss": 0.8395,
+      "step": 10039
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.077744999961166e-08,
+      "loss": 0.8162,
+      "step": 10040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.043490536463136e-08,
+      "loss": 0.7976,
+      "step": 10041
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.009332583281069e-08,
+      "loss": 0.8256,
+      "step": 10042
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.975271143731753e-08,
+      "loss": 0.8154,
+      "step": 10043
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941306221122545e-08,
+      "loss": 0.848,
+      "step": 10044
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9074378187510276e-08,
+      "loss": 0.8628,
+      "step": 10045
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.873665939906015e-08,
+      "loss": 0.7172,
+      "step": 10046
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.839990587866551e-08,
+      "loss": 0.9179,
+      "step": 10047
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.806411765902353e-08,
+      "loss": 0.8498,
+      "step": 10048
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7729294772740364e-08,
+      "loss": 0.8238,
+      "step": 10049
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.739543725232333e-08,
+      "loss": 0.8664,
+      "step": 10050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.706254513019094e-08,
+      "loss": 0.9075,
+      "step": 10051
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.673061843866623e-08,
+      "loss": 0.8714,
+      "step": 10052
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.639965720997787e-08,
+      "loss": 0.8266,
+      "step": 10053
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.606966147626125e-08,
+      "loss": 0.817,
+      "step": 10054
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.5740631269557423e-08,
+      "loss": 0.901,
+      "step": 10055
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.5412566621815266e-08,
+      "loss": 0.799,
+      "step": 10056
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.50854675648882e-08,
+      "loss": 0.8691,
+      "step": 10057
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.4759334130536355e-08,
+      "loss": 0.7877,
+      "step": 10058
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.443416635042775e-08,
+      "loss": 0.8294,
+      "step": 10059
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.41099642561338e-08,
+      "loss": 0.8299,
+      "step": 10060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.378672787913486e-08,
+      "loss": 0.8237,
+      "step": 10061
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3464457250814726e-08,
+      "loss": 0.8956,
+      "step": 10062
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.314315240246615e-08,
+      "loss": 0.7857,
+      "step": 10063
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.2822813365286386e-08,
+      "loss": 0.7893,
+      "step": 10064
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.250344017037945e-08,
+      "loss": 0.7827,
+      "step": 10065
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.218503284875609e-08,
+      "loss": 0.8843,
+      "step": 10066
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.1867591431332686e-08,
+      "loss": 0.8912,
+      "step": 10067
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.155111594893014e-08,
+      "loss": 0.8942,
+      "step": 10068
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.123560643227943e-08,
+      "loss": 0.8698,
+      "step": 10069
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.092106291201604e-08,
+      "loss": 0.9321,
+      "step": 10070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.060748541867888e-08,
+      "loss": 0.7985,
+      "step": 10071
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.029487398271693e-08,
+      "loss": 0.8163,
+      "step": 10072
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9983228634482574e-08,
+      "loss": 0.8559,
+      "step": 10073
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.967254940423716e-08,
+      "loss": 0.8813,
+      "step": 10074
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.936283632214656e-08,
+      "loss": 0.8654,
+      "step": 10075
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.905408941828338e-08,
+      "loss": 0.8048,
+      "step": 10076
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8746308722624756e-08,
+      "loss": 0.8237,
+      "step": 10077
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8439494265055674e-08,
+      "loss": 0.8487,
+      "step": 10078
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.813364607536786e-08,
+      "loss": 0.789,
+      "step": 10079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7828764183257545e-08,
+      "loss": 0.8927,
+      "step": 10080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.752484861832662e-08,
+      "loss": 0.773,
+      "step": 10081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.722189941008703e-08,
+      "loss": 0.8594,
+      "step": 10082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.691991658795303e-08,
+      "loss": 0.8235,
+      "step": 10083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.66189001812456e-08,
+      "loss": 0.8677,
+      "step": 10084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6318850219193576e-08,
+      "loss": 0.8626,
+      "step": 10085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6019766730930336e-08,
+      "loss": 0.8369,
+      "step": 10086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5721649745497084e-08,
+      "loss": 0.9142,
+      "step": 10087
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.542449929183845e-08,
+      "loss": 0.889,
+      "step": 10088
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.512831539880802e-08,
+      "loss": 0.8209,
+      "step": 10089
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.4833098095165005e-08,
+      "loss": 0.8901,
+      "step": 10090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.453884740957204e-08,
+      "loss": 0.8658,
+      "step": 10091
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.424556337060182e-08,
+      "loss": 0.8541,
+      "step": 10092
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.395324600673045e-08,
+      "loss": 0.8622,
+      "step": 10093
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.366189534634191e-08,
+      "loss": 0.7963,
+      "step": 10094
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.337151141772578e-08,
+      "loss": 0.9116,
+      "step": 10095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.308209424907506e-08,
+      "loss": 0.884,
+      "step": 10096
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.279364386849394e-08,
+      "loss": 0.8831,
+      "step": 10097
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.250616030399002e-08,
+      "loss": 0.7942,
+      "step": 10098
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.221964358347541e-08,
+      "loss": 0.8219,
+      "step": 10099
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1934093734771194e-08,
+      "loss": 0.8027,
+      "step": 10100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1649510785602976e-08,
+      "loss": 0.7832,
+      "step": 10101
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.13658947636042e-08,
+      "loss": 0.8389,
+      "step": 10102
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1083245696310614e-08,
+      "loss": 0.8632,
+      "step": 10103
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.080156361117027e-08,
+      "loss": 0.7235,
+      "step": 10104
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0520848535530176e-08,
+      "loss": 0.8337,
+      "step": 10105
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.024110049664853e-08,
+      "loss": 0.8918,
+      "step": 10106
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.996231952168805e-08,
+      "loss": 0.8169,
+      "step": 10107
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.9684505637718194e-08,
+      "loss": 0.9169,
+      "step": 10108
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.940765887171183e-08,
+      "loss": 0.8866,
+      "step": 10109
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.913177925055189e-08,
+      "loss": 0.7908,
+      "step": 10110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.885686680102474e-08,
+      "loss": 0.8318,
+      "step": 10111
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.858292154982457e-08,
+      "loss": 0.8533,
+      "step": 10112
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8309943523548995e-08,
+      "loss": 0.7974,
+      "step": 10113
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8037932748704596e-08,
+      "loss": 0.8919,
+      "step": 10114
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.776688925170247e-08,
+      "loss": 0.9736,
+      "step": 10115
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7496813058859325e-08,
+      "loss": 0.8325,
+      "step": 10116
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7227704196400874e-08,
+      "loss": 0.8139,
+      "step": 10117
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.695956269045509e-08,
+      "loss": 0.8877,
+      "step": 10118
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6692388567057813e-08,
+      "loss": 0.8222,
+      "step": 10119
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.642618185215163e-08,
+      "loss": 0.8004,
+      "step": 10120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6160942571583646e-08,
+      "loss": 0.8895,
+      "step": 10121
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.589667075110992e-08,
+      "loss": 0.8266,
+      "step": 10122
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5633366416386597e-08,
+      "loss": 0.8469,
+      "step": 10123
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.537102959298322e-08,
+      "loss": 0.752,
+      "step": 10124
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.510966030637164e-08,
+      "loss": 0.8348,
+      "step": 10125
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4849258581928224e-08,
+      "loss": 0.8524,
+      "step": 10126
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4589824444938305e-08,
+      "loss": 0.7792,
+      "step": 10127
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4331357920591726e-08,
+      "loss": 0.8284,
+      "step": 10128
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.407385903398508e-08,
+      "loss": 0.8394,
+      "step": 10129
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3817327810121706e-08,
+      "loss": 0.8638,
+      "step": 10130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.356176427390945e-08,
+      "loss": 0.8555,
+      "step": 10131
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3307168450160685e-08,
+      "loss": 0.8667,
+      "step": 10132
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.305354036359898e-08,
+      "loss": 0.8527,
+      "step": 10133
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.280088003885018e-08,
+      "loss": 0.7855,
+      "step": 10134
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.254918750044578e-08,
+      "loss": 0.8266,
+      "step": 10135
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.229846277282511e-08,
+      "loss": 0.8289,
+      "step": 10136
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2048705880333154e-08,
+      "loss": 0.8676,
+      "step": 10137
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1799916847220505e-08,
+      "loss": 0.8536,
+      "step": 10138
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.155209569764228e-08,
+      "loss": 0.8683,
+      "step": 10139
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.130524245566369e-08,
+      "loss": 0.8025,
+      "step": 10140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.10593571452511e-08,
+      "loss": 0.8436,
+      "step": 10141
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0814439790280984e-08,
+      "loss": 0.8323,
+      "step": 10142
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0570490414534304e-08,
+      "loss": 0.8131,
+      "step": 10143
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.032750904169546e-08,
+      "loss": 0.8961,
+      "step": 10144
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0085495695360015e-08,
+      "loss": 0.8817,
+      "step": 10145
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9844450399024725e-08,
+      "loss": 0.9213,
+      "step": 10146
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9604373176096434e-08,
+      "loss": 0.8871,
+      "step": 10147
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.936526404988427e-08,
+      "loss": 0.8875,
+      "step": 10148
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.912712304360521e-08,
+      "loss": 0.8543,
+      "step": 10149
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8889950180382985e-08,
+      "loss": 0.8617,
+      "step": 10150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8653745483244732e-08,
+      "loss": 0.7717,
+      "step": 10151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8418508975127655e-08,
+      "loss": 0.9163,
+      "step": 10152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8184240678871268e-08,
+      "loss": 0.8419,
+      "step": 10153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7950940617221812e-08,
+      "loss": 0.8387,
+      "step": 10154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.771860881283228e-08,
+      "loss": 0.8466,
+      "step": 10155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7487245288261298e-08,
+      "loss": 0.7534,
+      "step": 10156
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.725685006597534e-08,
+      "loss": 0.8206,
+      "step": 10157
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.702742316834206e-08,
+      "loss": 0.8164,
+      "step": 10158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6798964617641422e-08,
+      "loss": 0.8999,
+      "step": 10159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.657147443605457e-08,
+      "loss": 0.8428,
+      "step": 10160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6344952645669386e-08,
+      "loss": 0.8718,
+      "step": 10161
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6119399268480505e-08,
+      "loss": 0.9276,
+      "step": 10162
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5894814326390407e-08,
+      "loss": 0.8838,
+      "step": 10163
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5671197841203867e-08,
+      "loss": 0.869,
+      "step": 10164
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.544854983463463e-08,
+      "loss": 0.8806,
+      "step": 10165
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5226870328299845e-08,
+      "loss": 0.8387,
+      "step": 10166
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5006159343724524e-08,
+      "loss": 0.907,
+      "step": 10167
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4786416902339293e-08,
+      "loss": 0.8311,
+      "step": 10168
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4567643025481534e-08,
+      "loss": 0.9084,
+      "step": 10169
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4349837734390923e-08,
+      "loss": 0.7889,
+      "step": 10170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.413300105021832e-08,
+      "loss": 0.9237,
+      "step": 10171
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3917132994016878e-08,
+      "loss": 0.7932,
+      "step": 10172
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3702233586745395e-08,
+      "loss": 0.8223,
+      "step": 10173
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3488302849272726e-08,
+      "loss": 0.8523,
+      "step": 10174
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3275340802368927e-08,
+      "loss": 0.8488,
+      "step": 10175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3063347466713014e-08,
+      "loss": 0.861,
+      "step": 10176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.285232286288741e-08,
+      "loss": 0.8728,
+      "step": 10177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.264226701138461e-08,
+      "loss": 0.803,
+      "step": 10178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2433179932597192e-08,
+      "loss": 0.8533,
+      "step": 10179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2225061646830025e-08,
+      "loss": 0.8793,
+      "step": 10180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.201791217428917e-08,
+      "loss": 0.8958,
+      "step": 10181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.181173153508853e-08,
+      "loss": 0.889,
+      "step": 10182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1606519749247657e-08,
+      "loss": 0.8754,
+      "step": 10183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1402276836691715e-08,
+      "loss": 0.848,
+      "step": 10184
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1199002817252623e-08,
+      "loss": 0.8905,
+      "step": 10185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0996697710666812e-08,
+      "loss": 0.8431,
+      "step": 10186
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0795361536578575e-08,
+      "loss": 0.9017,
+      "step": 10187
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.059499431453671e-08,
+      "loss": 0.8322,
+      "step": 10188
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.039559606399677e-08,
+      "loss": 0.8976,
+      "step": 10189
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0197166804317713e-08,
+      "loss": 0.8315,
+      "step": 10190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.999970655476968e-08,
+      "loss": 0.8506,
+      "step": 10191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9803215334522895e-08,
+      "loss": 0.9697,
+      "step": 10192
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9607693162657647e-08,
+      "loss": 0.7711,
+      "step": 10193
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9413140058156533e-08,
+      "loss": 0.8201,
+      "step": 10194
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9219556039913324e-08,
+      "loss": 0.9064,
+      "step": 10195
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9026941126721876e-08,
+      "loss": 0.8742,
+      "step": 10196
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8835295337283898e-08,
+      "loss": 0.839,
+      "step": 10197
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.864461869021117e-08,
+      "loss": 0.9108,
+      "step": 10198
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8454911204014437e-08,
+      "loss": 0.8878,
+      "step": 10199
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.826617289711563e-08,
+      "loss": 0.9058,
+      "step": 10200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8078403787838984e-08,
+      "loss": 0.837,
+      "step": 10201
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7891603894418797e-08,
+      "loss": 0.8649,
+      "step": 10202
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7705773234990564e-08,
+      "loss": 0.881,
+      "step": 10203
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7520911827598742e-08,
+      "loss": 0.8704,
+      "step": 10204
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7337019690193413e-08,
+      "loss": 0.8726,
+      "step": 10205
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7154096840629186e-08,
+      "loss": 0.8395,
+      "step": 10206
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.697214329666852e-08,
+      "loss": 0.8552,
+      "step": 10207
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.679115907597617e-08,
+      "loss": 0.8552,
+      "step": 10208
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6611144196128082e-08,
+      "loss": 0.8463,
+      "step": 10209
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6432098674600272e-08,
+      "loss": 0.7723,
+      "step": 10210
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6254022528779946e-08,
+      "loss": 0.8122,
+      "step": 10211
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6076915775956604e-08,
+      "loss": 0.8058,
+      "step": 10212
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.5900778433327603e-08,
+      "loss": 0.9137,
+      "step": 10213
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5725610517994816e-08,
+      "loss": 0.8104,
+      "step": 10214
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5551412046966862e-08,
+      "loss": 0.8962,
+      "step": 10215
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.537818303715688e-08,
+      "loss": 0.8235,
+      "step": 10216
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5205923505385857e-08,
+      "loss": 0.8881,
+      "step": 10217
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.50346334683793e-08,
+      "loss": 0.843,
+      "step": 10218
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4864312942769465e-08,
+      "loss": 0.8716,
+      "step": 10219
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4694961945093122e-08,
+      "loss": 0.8737,
+      "step": 10220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4526580491794895e-08,
+      "loss": 0.8161,
+      "step": 10221
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4359168599223926e-08,
+      "loss": 0.8357,
+      "step": 10222
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4192726283634995e-08,
+      "loss": 0.9031,
+      "step": 10223
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4027253561188502e-08,
+      "loss": 0.9079,
+      "step": 10224
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.386275044795271e-08,
+      "loss": 0.8714,
+      "step": 10225
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3699216959899287e-08,
+      "loss": 0.7876,
+      "step": 10226
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.353665311290886e-08,
+      "loss": 0.8685,
+      "step": 10227
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.337505892276325e-08,
+      "loss": 0.8073,
+      "step": 10228
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3214434405154352e-08,
+      "loss": 0.8413,
+      "step": 10229
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3054779575677468e-08,
+      "loss": 0.8799,
+      "step": 10230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.289609444983575e-08,
+      "loss": 0.9221,
+      "step": 10231
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2738379043035765e-08,
+      "loss": 0.8783,
+      "step": 10232
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2581633370593038e-08,
+      "loss": 0.8839,
+      "step": 10233
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2425857447725397e-08,
+      "loss": 0.82,
+      "step": 10234
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2271051289558522e-08,
+      "loss": 0.8172,
+      "step": 10235
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.211721491112372e-08,
+      "loss": 0.7806,
+      "step": 10236
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1964348327359044e-08,
+      "loss": 0.8638,
+      "step": 10237
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1812451553107063e-08,
+      "loss": 0.8396,
+      "step": 10238
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.166152460311487e-08,
+      "loss": 0.9973,
+      "step": 10239
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1511567492038522e-08,
+      "loss": 0.8437,
+      "step": 10240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1362580234438592e-08,
+      "loss": 0.8065,
+      "step": 10241
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1214562844781285e-08,
+      "loss": 0.8565,
+      "step": 10242
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1067515337437329e-08,
+      "loss": 0.8581,
+      "step": 10243
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0921437726686413e-08,
+      "loss": 0.802,
+      "step": 10244
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0776330026711635e-08,
+      "loss": 0.8669,
+      "step": 10245
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0632192251601726e-08,
+      "loss": 0.8185,
+      "step": 10246
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0489024415352156e-08,
+      "loss": 0.8803,
+      "step": 10247
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0346826531865139e-08,
+      "loss": 0.8616,
+      "step": 10248
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.02055986149463e-08,
+      "loss": 0.8597,
+      "step": 10249
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.006534067831022e-08,
+      "loss": 0.776,
+      "step": 10250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.926052735574898e-09,
+      "loss": 0.8363,
+      "step": 10251
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.78773480026396e-09,
+      "loss": 0.8628,
+      "step": 10252
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.650386885808882e-09,
+      "loss": 0.9016,
+      "step": 10253
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.514009005543445e-09,
+      "loss": 0.8456,
+      "step": 10254
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.37860117271261e-09,
+      "loss": 0.8172,
+      "step": 10255
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.244163400462525e-09,
+      "loss": 0.7542,
+      "step": 10256
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.110695701847194e-09,
+      "loss": 0.8293,
+      "step": 10257
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.97819808982403e-09,
+      "loss": 0.8832,
+      "step": 10258
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.84667057726163e-09,
+      "loss": 0.8743,
+      "step": 10259
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.716113176927554e-09,
+      "loss": 0.8729,
+      "step": 10260
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.586525901500552e-09,
+      "loss": 0.8289,
+      "step": 10261
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.457908763562783e-09,
+      "loss": 0.9661,
+      "step": 10262
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.330261775603143e-09,
+      "loss": 0.8542,
+      "step": 10263
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.203584950013942e-09,
+      "loss": 0.929,
+      "step": 10264
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.077878299097563e-09,
+      "loss": 0.8608,
+      "step": 10265
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.953141835057576e-09,
+      "loss": 0.8224,
+      "step": 10266
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.829375570006514e-09,
+      "loss": 1.0432,
+      "step": 10267
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.706579515962542e-09,
+      "loss": 0.8773,
+      "step": 10268
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.584753684847235e-09,
+      "loss": 0.9054,
+      "step": 10269
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.463898088490019e-09,
+      "loss": 0.8371,
+      "step": 10270
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.34401273862595e-09,
+      "loss": 0.8395,
+      "step": 10271
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.225097646895718e-09,
+      "loss": 0.9443,
+      "step": 10272
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.107152824845642e-09,
+      "loss": 0.8427,
+      "step": 10273
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.990178283927673e-09,
+      "loss": 0.9105,
+      "step": 10274
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.8741740354993924e-09,
+      "loss": 0.8817,
+      "step": 10275
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.759140090824012e-09,
+      "loss": 0.8499,
+      "step": 10276
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.645076461072597e-09,
+      "loss": 0.8926,
+      "step": 10277
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.531983157318511e-09,
+      "loss": 0.8606,
+      "step": 10278
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.4198601905440805e-09,
+      "loss": 0.8689,
+      "step": 10279
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.308707571636152e-09,
+      "loss": 0.8562,
+      "step": 10280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.198525311387205e-09,
+      "loss": 0.8749,
+      "step": 10281
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.089313420494236e-09,
+      "loss": 0.8416,
+      "step": 10282
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.981071909563208e-09,
+      "loss": 0.9233,
+      "step": 10283
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.87380078910349e-09,
+      "loss": 0.8215,
+      "step": 10284
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.767500069531195e-09,
+      "loss": 0.8539,
+      "step": 10285
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.662169761165848e-09,
+      "loss": 0.9093,
+      "step": 10286
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.557809874237041e-09,
+      "loss": 0.8777,
+      "step": 10287
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4544204188777815e-09,
+      "loss": 0.9363,
+      "step": 10288
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.352001405124485e-09,
+      "loss": 0.8902,
+      "step": 10289
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.25055284292475e-09,
+      "loss": 0.7995,
+      "step": 10290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.150074742126254e-09,
+      "loss": 0.8068,
+      "step": 10291
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.050567112487858e-09,
+      "loss": 0.8271,
+      "step": 10292
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9520299636685036e-09,
+      "loss": 0.8358,
+      "step": 10293
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.85446330523831e-09,
+      "loss": 0.865,
+      "step": 10294
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.757867146669703e-09,
+      "loss": 0.8064,
+      "step": 10295
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.662241497341846e-09,
+      "loss": 0.8414,
+      "step": 10296
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.567586366539534e-09,
+      "loss": 0.8136,
+      "step": 10297
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.473901763454302e-09,
+      "loss": 0.8482,
+      "step": 10298
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.38118769718221e-09,
+      "loss": 0.8166,
+      "step": 10299
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.289444176724944e-09,
+      "loss": 0.8465,
+      "step": 10300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.198671210990934e-09,
+      "loss": 0.8275,
+      "step": 10301
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.108868808794242e-09,
+      "loss": 0.8636,
+      "step": 10302
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.020036978854558e-09,
+      "loss": 0.8705,
+      "step": 10303
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.932175729797205e-09,
+      "loss": 0.8534,
+      "step": 10304
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.845285070153138e-09,
+      "loss": 0.8453,
+      "step": 10305
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.759365008357829e-09,
+      "loss": 0.8456,
+      "step": 10306
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.6744155527557167e-09,
+      "loss": 0.7976,
+      "step": 10307
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.590436711594647e-09,
+      "loss": 0.7616,
+      "step": 10308
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.507428493029208e-09,
+      "loss": 0.864,
+      "step": 10309
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4253909051173985e-09,
+      "loss": 0.8522,
+      "step": 10310
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.344323955827289e-09,
+      "loss": 0.8517,
+      "step": 10311
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.2642276530281404e-09,
+      "loss": 0.832,
+      "step": 10312
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.185102004498175e-09,
+      "loss": 0.9556,
+      "step": 10313
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1069470179201365e-09,
+      "loss": 0.8704,
+      "step": 10314
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.029762700882399e-09,
+      "loss": 0.8527,
+      "step": 10315
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9535490608789687e-09,
+      "loss": 0.9379,
+      "step": 10316
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.878306105311701e-09,
+      "loss": 0.9387,
+      "step": 10317
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8040338414847545e-09,
+      "loss": 0.8876,
+      "step": 10318
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7307322766101373e-09,
+      "loss": 0.8528,
+      "step": 10319
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6584014178054894e-09,
+      "loss": 0.8806,
+      "step": 10320
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5870412720929715e-09,
+      "loss": 0.8316,
+      "step": 10321
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.516651846403706e-09,
+      "loss": 0.8352,
+      "step": 10322
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.447233147570005e-09,
+      "loss": 0.8587,
+      "step": 10323
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.378785182333143e-09,
+      "loss": 0.8993,
+      "step": 10324
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3113079573389153e-09,
+      "loss": 0.8628,
+      "step": 10325
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2448014791398574e-09,
+      "loss": 0.9178,
+      "step": 10326
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1792657541930273e-09,
+      "loss": 0.8701,
+      "step": 10327
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1147007888622227e-09,
+      "loss": 0.8739,
+      "step": 10328
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0511065894157634e-09,
+      "loss": 0.869,
+      "step": 10329
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9884831620287094e-09,
+      "loss": 0.8129,
+      "step": 10330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.926830512781752e-09,
+      "loss": 0.8129,
+      "step": 10331
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8661486476612144e-09,
+      "loss": 0.8307,
+      "step": 10332
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8064375725590498e-09,
+      "loss": 0.9352,
+      "step": 10333
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.747697293272843e-09,
+      "loss": 0.9717,
+      "step": 10334
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.68992781550692e-09,
+      "loss": 0.9312,
+      "step": 10335
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.633129144870127e-09,
+      "loss": 0.8405,
+      "step": 10336
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5773012868769422e-09,
+      "loss": 0.8406,
+      "step": 10337
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5224442469474743e-09,
+      "loss": 0.8417,
+      "step": 10338
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4685580304107939e-09,
+      "loss": 0.8457,
+      "step": 10339
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4156426424960513e-09,
+      "loss": 0.8699,
+      "step": 10340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.363698088343579e-09,
+      "loss": 0.8922,
+      "step": 10341
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3127243729949003e-09,
+      "loss": 0.8582,
+      "step": 10342
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2627215014016092e-09,
+      "loss": 0.8384,
+      "step": 10343
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2136894784176011e-09,
+      "loss": 0.8158,
+      "step": 10344
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1656283088035126e-09,
+      "loss": 0.8821,
+      "step": 10345
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1185379972256105e-09,
+      "loss": 0.8262,
+      "step": 10346
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0724185482580141e-09,
+      "loss": 0.8549,
+      "step": 10347
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0272699663782525e-09,
+      "loss": 0.8584,
+      "step": 10348
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.830922559683764e-10,
+      "loss": 0.8858,
+      "step": 10349
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.398854213193976e-10,
+      "loss": 0.783,
+      "step": 10350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.976494666268487e-10,
+      "loss": 0.8267,
+      "step": 10351
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.563843959907836e-10,
+      "loss": 0.8832,
+      "step": 10352
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.160902134191074e-10,
+      "loss": 0.8703,
+      "step": 10353
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.767669228231356e-10,
+      "loss": 0.8452,
+      "step": 10354
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.384145280209254e-10,
+      "loss": 0.8625,
+      "step": 10355
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.010330327372749e-10,
+      "loss": 0.8469,
+      "step": 10356
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.646224406003932e-10,
+      "loss": 0.8658,
+      "step": 10357
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.291827551474505e-10,
+      "loss": 0.8361,
+      "step": 10358
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.947139798190282e-10,
+      "loss": 0.7933,
+      "step": 10359
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.612161179613385e-10,
+      "loss": 0.8236,
+      "step": 10360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.286891728273346e-10,
+      "loss": 0.8868,
+      "step": 10361
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.971331475756013e-10,
+      "loss": 0.8683,
+      "step": 10362
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.665480452692439e-10,
+      "loss": 0.8674,
+      "step": 10363
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.369338688781088e-10,
+      "loss": 0.8958,
+      "step": 10364
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0829062127878407e-10,
+      "loss": 0.8485,
+      "step": 10365
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.8061830525126797e-10,
+      "loss": 0.8439,
+      "step": 10366
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.5391692348341057e-10,
+      "loss": 0.8456,
+      "step": 10367
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.2818647856647234e-10,
+      "loss": 0.9015,
+      "step": 10368
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.034269730006756e-10,
+      "loss": 0.8244,
+      "step": 10369
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.796384091885429e-10,
+      "loss": 0.8635,
+      "step": 10370
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.568207894404484e-10,
+      "loss": 0.8564,
+      "step": 10371
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3497411597128705e-10,
+      "loss": 0.8733,
+      "step": 10372
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1409839090380525e-10,
+      "loss": 0.8868,
+      "step": 10373
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9419361626416e-10,
+      "loss": 0.8982,
+      "step": 10374
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.752597939841394e-10,
+      "loss": 0.8411,
+      "step": 10375
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5729692590338296e-10,
+      "loss": 0.8658,
+      "step": 10376
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4030501376494087e-10,
+      "loss": 0.848,
+      "step": 10377
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2428405921971476e-10,
+      "loss": 0.8779,
+      "step": 10378
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0923406382423729e-10,
+      "loss": 0.7921,
+      "step": 10379
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.515502903734153e-11,
+      "loss": 0.912,
+      "step": 10380
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.204695622682223e-11,
+      "loss": 0.9299,
+      "step": 10381
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.990984666610523e-11,
+      "loss": 0.8641,
+      "step": 10382
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.874370153424736e-11,
+      "loss": 0.8458,
+      "step": 10383
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.854852191371606e-11,
+      "loss": 0.8672,
+      "step": 10384
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.932430879594051e-11,
+      "loss": 0.8358,
+      "step": 10385
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.1071063075760466e-11,
+      "loss": 0.8368,
+      "step": 10386
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3788785554756942e-11,
+      "loss": 0.8927,
+      "step": 10387
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7477476940142013e-11,
+      "loss": 0.7978,
+      "step": 10388
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2137137843648562e-11,
+      "loss": 0.8419,
+      "step": 10389
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.767768785971185e-12,
+      "loss": 0.8325,
+      "step": 10390
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3693701889946375e-12,
+      "loss": 0.849,
+      "step": 10391
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9419423846755993e-12,
+      "loss": 0.8755,
+      "step": 10392
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.854856083813531e-13,
+      "loss": 0.8205,
+      "step": 10393
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.8049,
+      "step": 10394
+    },
+    {
+      "epoch": 2.0,
+      "step": 10394,
+      "total_flos": 7544441038274560.0,
+      "train_loss": 0.9752291489821892,
+      "train_runtime": 95947.0807,
+      "train_samples_per_second": 13.868,
+      "train_steps_per_second": 0.108
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10394,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 50000,
+  "total_flos": 7544441038274560.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}