diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31195 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998556512534282,
+  "global_step": 5195,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 1.2913,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 1.3433,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 1.2906,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 1.32,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 1.3552,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.2968,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 1.3286,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 1.323,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 1.2886,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 1.2932,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 1.3063,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.2905,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.1951,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 1.1038,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.1168,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.1697,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 1.1439,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.0741,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.0584,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.1403,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.1109,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.1065,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 1.0922,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.1068,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 1.0774,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.088,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.0601,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.0948,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.0305,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.9492,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 1.0178,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 0.9447,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.0362,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 0.9869,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 0.9962,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.016,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 0.9641,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.0203,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.9719,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.0445,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 0.916,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 0.3084,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 0.9646,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 0.28,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 0.9775,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 0.9631,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 0.9577,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.9622,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 0.9614,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 0.3078,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 0.9512,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.9823,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 0.9079,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.9159,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 0.9701,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 0.9828,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 0.979,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 0.9499,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 1.0214,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.2869,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 0.9348,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 0.9633,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 0.9193,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 0.9363,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.0191,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.9213,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 0.9621,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 0.2692,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.9022,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.9809,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 0.9585,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.9332,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 0.9416,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 0.8758,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.8591,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.9185,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 0.9142,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9277,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 0.9568,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.9516,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 0.9373,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 0.9226,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0641025641025643e-05,
+      "loss": 0.937,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.9122,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 1.0271,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 0.8925,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.9614,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.883,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 0.9642,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.8748,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.8847,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 0.9486,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 0.9596,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 0.9492,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 0.9593,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.902,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 0.2789,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 0.9874,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 0.8777,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 0.9438,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 0.9349,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.8988,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 0.8928,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.2999,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 0.9377,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 0.9427,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3717948717948718e-05,
+      "loss": 0.9646,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.8997,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 1.018,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 0.8681,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 0.93,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 0.9355,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 0.9304,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 1.0162,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 0.9096,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 0.8894,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.915,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 0.9491,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5256410256410257e-05,
+      "loss": 0.9369,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.8817,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 0.9673,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 0.9454,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 0.9256,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.9152,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 0.9471,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.8621,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 0.9681,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.9122,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 0.9228,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.9367,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 0.9531,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.9053,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 1.0133,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 0.9233,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.9394,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 0.9286,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7564102564102566e-05,
+      "loss": 0.3205,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.8666,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 0.8762,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.9171,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 0.9932,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 0.8365,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.8791,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.9077,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8589743589743593e-05,
+      "loss": 0.9109,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 0.8796,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 0.9272,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 0.9171,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9102564102564106e-05,
+      "loss": 0.315,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.9012,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 0.9333,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 0.908,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 0.9342,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 0.9029,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9871794871794873e-05,
+      "loss": 0.951,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.2841,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999805651578e-05,
+      "loss": 0.9264,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992226063874e-05,
+      "loss": 0.9507,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999982508646547e-05,
+      "loss": 0.9281,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999968904267578e-05,
+      "loss": 0.979,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995141293225e-05,
+      "loss": 0.9572,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993003464737e-05,
+      "loss": 0.8116,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999904769421242e-05,
+      "loss": 0.9516,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987561726369e-05,
+      "loss": 0.9192,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984257818604e-05,
+      "loss": 0.923,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999805652201144e-05,
+      "loss": 0.8894,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999976483932334e-05,
+      "loss": 0.9278,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999720139568508e-05,
+      "loss": 0.9314,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999671552954013e-05,
+      "loss": 0.908,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999961907949874e-05,
+      "loss": 0.2894,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956271922309e-05,
+      "loss": 0.9284,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999502472148968e-05,
+      "loss": 0.9361,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999943833829979e-05,
+      "loss": 0.974,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999937031770049e-05,
+      "loss": 0.3087,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99992984103775e-05,
+      "loss": 0.9613,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999222616358778e-05,
+      "loss": 0.8356,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999142935673778e-05,
+      "loss": 0.9052,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999059368353474e-05,
+      "loss": 0.9393,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999897191443035e-05,
+      "loss": 0.9045,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998880573938396e-05,
+      "loss": 0.9282,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998785346913123e-05,
+      "loss": 0.9351,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998686233391537e-05,
+      "loss": 0.9222,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998583233412162e-05,
+      "loss": 0.8992,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998476347015042e-05,
+      "loss": 0.9613,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999836557424172e-05,
+      "loss": 0.9572,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998250915135255e-05,
+      "loss": 0.8404,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998132369740207e-05,
+      "loss": 0.8912,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999800993810266e-05,
+      "loss": 0.8859,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997883620270203e-05,
+      "loss": 0.9257,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997753416291935e-05,
+      "loss": 0.9596,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997619326218465e-05,
+      "loss": 0.9513,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999748135010191e-05,
+      "loss": 0.8456,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999733948799591e-05,
+      "loss": 0.9247,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99971937399556e-05,
+      "loss": 0.9945,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999704410603763e-05,
+      "loss": 0.8833,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996890586300168e-05,
+      "loss": 0.9401,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999673318080288e-05,
+      "loss": 0.9511,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996571889606956e-05,
+      "loss": 0.9325,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999640671277508e-05,
+      "loss": 0.9466,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999623765037147e-05,
+      "loss": 0.901,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999606470246183e-05,
+      "loss": 0.8805,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995887869113382e-05,
+      "loss": 0.9104,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999570715039487e-05,
+      "loss": 0.9137,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995522546376528e-05,
+      "loss": 0.8911,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995334057130117e-05,
+      "loss": 0.9361,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995141682728904e-05,
+      "loss": 0.7638,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999494542324766e-05,
+      "loss": 0.8543,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994745278762676e-05,
+      "loss": 0.9376,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999454124935174e-05,
+      "loss": 0.8787,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994333335094162e-05,
+      "loss": 0.9739,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994121536070757e-05,
+      "loss": 0.9149,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999390585236385e-05,
+      "loss": 0.9457,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993686284057283e-05,
+      "loss": 0.9645,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993462831236386e-05,
+      "loss": 0.8828,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999323549398803e-05,
+      "loss": 0.8627,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993004272400576e-05,
+      "loss": 0.8947,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992769166563894e-05,
+      "loss": 0.9422,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992530176569372e-05,
+      "loss": 0.9379,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999228730250991e-05,
+      "loss": 0.8761,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992040544479904e-05,
+      "loss": 0.963,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991789902575273e-05,
+      "loss": 0.8894,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991535376893435e-05,
+      "loss": 0.8963,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991276967533332e-05,
+      "loss": 0.8797,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99910146745954e-05,
+      "loss": 0.9282,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990748498181597e-05,
+      "loss": 0.935,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999047843839538e-05,
+      "loss": 0.9343,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999020449534172e-05,
+      "loss": 0.8751,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989926669127105e-05,
+      "loss": 0.8876,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989644959859514e-05,
+      "loss": 0.9767,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989359367648457e-05,
+      "loss": 0.9132,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998906989260494e-05,
+      "loss": 0.9128,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988776534841477e-05,
+      "loss": 0.3259,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988479294472098e-05,
+      "loss": 0.293,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988178171612337e-05,
+      "loss": 0.9044,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987873166379245e-05,
+      "loss": 0.9751,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998756427889137e-05,
+      "loss": 0.9143,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998725150926878e-05,
+      "loss": 0.8403,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986934857633053e-05,
+      "loss": 0.9007,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986614324107255e-05,
+      "loss": 1.0079,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986289908815986e-05,
+      "loss": 0.9607,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998596161188535e-05,
+      "loss": 0.931,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985629433442944e-05,
+      "loss": 0.8753,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985293373617892e-05,
+      "loss": 0.8771,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998495343254082e-05,
+      "loss": 0.9107,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984609610343857e-05,
+      "loss": 1.0345,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984261907160653e-05,
+      "loss": 0.9469,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983910323126346e-05,
+      "loss": 0.9317,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998355485837761e-05,
+      "loss": 0.9074,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983195513052605e-05,
+      "loss": 0.3702,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998283228729101e-05,
+      "loss": 0.9099,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982465181234007e-05,
+      "loss": 0.2994,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982094195024294e-05,
+      "loss": 0.9316,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981719328806065e-05,
+      "loss": 0.9135,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981340582725038e-05,
+      "loss": 0.9015,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980957956928425e-05,
+      "loss": 0.8666,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998057145156495e-05,
+      "loss": 0.885,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998018106678485e-05,
+      "loss": 0.3806,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979786802739863e-05,
+      "loss": 0.8856,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979388659583246e-05,
+      "loss": 0.9032,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978986637469747e-05,
+      "loss": 0.9587,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978580736555633e-05,
+      "loss": 0.9307,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978170956998677e-05,
+      "loss": 0.3478,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977757298958164e-05,
+      "loss": 0.9322,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977339762594868e-05,
+      "loss": 0.87,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99769183480711e-05,
+      "loss": 0.9734,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976493055550655e-05,
+      "loss": 0.8596,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997606388519884e-05,
+      "loss": 0.8997,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997563083718248e-05,
+      "loss": 0.9214,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997519391166989e-05,
+      "loss": 0.9175,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997475310883091e-05,
+      "loss": 0.9607,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997430842883688e-05,
+      "loss": 0.9357,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973859871860635e-05,
+      "loss": 0.8822,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973407438076533e-05,
+      "loss": 0.8914,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972951127660434e-05,
+      "loss": 0.9107,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972490940789708e-05,
+      "loss": 0.9553,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972026877643223e-05,
+      "loss": 0.8554,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997155893840136e-05,
+      "loss": 0.8769,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971087123246008e-05,
+      "loss": 0.9023,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970611432360556e-05,
+      "loss": 0.911,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997013186592991e-05,
+      "loss": 0.8698,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969648424140467e-05,
+      "loss": 0.9575,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969161107180146e-05,
+      "loss": 0.9165,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968669915238364e-05,
+      "loss": 0.9123,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968174848506044e-05,
+      "loss": 0.9297,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967675907175617e-05,
+      "loss": 0.9021,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967173091441022e-05,
+      "loss": 0.9512,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966666401497707e-05,
+      "loss": 0.9206,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966155837542606e-05,
+      "loss": 0.8697,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965641399774187e-05,
+      "loss": 0.9526,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965123088392407e-05,
+      "loss": 0.8828,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964600903598724e-05,
+      "loss": 0.3344,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964074845596125e-05,
+      "loss": 0.9568,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963544914589073e-05,
+      "loss": 0.9382,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996301111078356e-05,
+      "loss": 0.8241,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962473434387068e-05,
+      "loss": 0.854,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996193188560859e-05,
+      "loss": 0.8357,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961386464658626e-05,
+      "loss": 0.8944,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960837171749183e-05,
+      "loss": 0.9027,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960284007093764e-05,
+      "loss": 0.8741,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995972697090738e-05,
+      "loss": 0.9353,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959166063406563e-05,
+      "loss": 0.9131,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995860128480932e-05,
+      "loss": 0.9723,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958032635335187e-05,
+      "loss": 0.8623,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957460115205192e-05,
+      "loss": 0.9697,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995688372464188e-05,
+      "loss": 0.9387,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956303463869285e-05,
+      "loss": 0.9568,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995571933311295e-05,
+      "loss": 0.8719,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955131332599934e-05,
+      "loss": 0.8934,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995453946255878e-05,
+      "loss": 0.9001,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953943723219553e-05,
+      "loss": 0.8616,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953344114813816e-05,
+      "loss": 0.8761,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995274063757463e-05,
+      "loss": 0.8812,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995213329173657e-05,
+      "loss": 0.8585,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995152207753571e-05,
+      "loss": 0.9342,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950906995209617e-05,
+      "loss": 0.9476,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995028804499738e-05,
+      "loss": 0.8864,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949665227139583e-05,
+      "loss": 1.0203,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949038541878312e-05,
+      "loss": 0.8303,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948407989457156e-05,
+      "loss": 0.9256,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994777357012121e-05,
+      "loss": 0.9767,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994713528411707e-05,
+      "loss": 0.8073,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946493131692837e-05,
+      "loss": 0.9609,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945847113098112e-05,
+      "loss": 0.9068,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945197228584006e-05,
+      "loss": 0.8708,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994454347840312e-05,
+      "loss": 0.8918,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943885862809566e-05,
+      "loss": 0.9165,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994322438205896e-05,
+      "loss": 0.9009,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942559036408413e-05,
+      "loss": 0.885,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941889826116546e-05,
+      "loss": 0.9571,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941216751443485e-05,
+      "loss": 0.8895,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994053981265084e-05,
+      "loss": 0.8453,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993985901000174e-05,
+      "loss": 0.9243,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939174343760813e-05,
+      "loss": 0.8457,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938485814194182e-05,
+      "loss": 0.9396,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937793421569482e-05,
+      "loss": 0.9294,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993709716615584e-05,
+      "loss": 0.8868,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993639704822389e-05,
+      "loss": 0.947,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935693068045765e-05,
+      "loss": 0.8514,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99349852258951e-05,
+      "loss": 0.9022,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934273522047028e-05,
+      "loss": 0.8822,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933557956778192e-05,
+      "loss": 0.9645,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932838530366726e-05,
+      "loss": 0.8809,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993211524309227e-05,
+      "loss": 0.8509,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931388095235964e-05,
+      "loss": 0.8719,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930657087080446e-05,
+      "loss": 0.9455,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929922218909855e-05,
+      "loss": 0.9122,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929183491009837e-05,
+      "loss": 0.9019,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928440903667532e-05,
+      "loss": 0.3247,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992769445717158e-05,
+      "loss": 0.9186,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926944151812124e-05,
+      "loss": 0.8499,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99261899878808e-05,
+      "loss": 0.9008,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925431965670752e-05,
+      "loss": 0.8849,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924670085476626e-05,
+      "loss": 0.9112,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923904347594558e-05,
+      "loss": 0.9121,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923134752322187e-05,
+      "loss": 0.8222,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922361299958654e-05,
+      "loss": 0.8896,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99215839908046e-05,
+      "loss": 0.3359,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920802825162153e-05,
+      "loss": 0.9427,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992001780333496e-05,
+      "loss": 0.8873,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919228925628152e-05,
+      "loss": 0.9106,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918436192348362e-05,
+      "loss": 0.8563,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991763960380373e-05,
+      "loss": 0.9384,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916839160303878e-05,
+      "loss": 0.3312,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916034862159943e-05,
+      "loss": 0.8789,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991522670968455e-05,
+      "loss": 0.8674,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914414703191823e-05,
+      "loss": 0.8716,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913598842997394e-05,
+      "loss": 0.8858,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912779129418377e-05,
+      "loss": 0.8797,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991195556277339e-05,
+      "loss": 0.8484,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991112814338257e-05,
+      "loss": 0.3082,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910296871567508e-05,
+      "loss": 0.9341,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990946174765133e-05,
+      "loss": 0.8604,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990862277195864e-05,
+      "loss": 0.9028,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907779944815554e-05,
+      "loss": 0.9712,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906933266549665e-05,
+      "loss": 0.9774,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906082737490086e-05,
+      "loss": 0.929,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905228357967403e-05,
+      "loss": 0.9302,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990437012831372e-05,
+      "loss": 0.9247,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903508048862626e-05,
+      "loss": 0.3477,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902642119949203e-05,
+      "loss": 0.9108,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901772341910043e-05,
+      "loss": 0.8436,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900898715083218e-05,
+      "loss": 0.9192,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900021239808312e-05,
+      "loss": 0.8393,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989913991642639e-05,
+      "loss": 0.8506,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898254745280024e-05,
+      "loss": 0.9376,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897365726713277e-05,
+      "loss": 0.901,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896472861071704e-05,
+      "loss": 0.9465,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989557614870236e-05,
+      "loss": 0.9062,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98946755899538e-05,
+      "loss": 0.9174,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989377118517606e-05,
+      "loss": 0.8823,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892862934720685e-05,
+      "loss": 0.8326,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891950838940708e-05,
+      "loss": 0.9054,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891034898190657e-05,
+      "loss": 0.9106,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890115112826556e-05,
+      "loss": 0.9272,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988919148320592e-05,
+      "loss": 0.9096,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888264009687765e-05,
+      "loss": 0.8819,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887332692632596e-05,
+      "loss": 0.8889,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988639753240241e-05,
+      "loss": 0.8676,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885458529360704e-05,
+      "loss": 0.8459,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884515683872464e-05,
+      "loss": 0.9279,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988356899630417e-05,
+      "loss": 0.9423,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882618467023804e-05,
+      "loss": 0.92,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988166409640082e-05,
+      "loss": 0.8151,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988070588480619e-05,
+      "loss": 0.8359,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879743832612367e-05,
+      "loss": 0.9127,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878777940193294e-05,
+      "loss": 0.8437,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877808207924406e-05,
+      "loss": 0.8876,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876834636182644e-05,
+      "loss": 0.8831,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987585722534643e-05,
+      "loss": 0.9067,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874875975795677e-05,
+      "loss": 0.8392,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873890887911796e-05,
+      "loss": 0.9112,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872901962077687e-05,
+      "loss": 0.9402,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871909198677744e-05,
+      "loss": 0.8627,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870912598097847e-05,
+      "loss": 0.8842,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986991216072538e-05,
+      "loss": 0.8708,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868907886949197e-05,
+      "loss": 0.8399,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867899777159665e-05,
+      "loss": 0.8706,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866887831748633e-05,
+      "loss": 0.8898,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865872051109435e-05,
+      "loss": 0.9232,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986485243563691e-05,
+      "loss": 0.8878,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863828985727373e-05,
+      "loss": 0.9057,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862801701778637e-05,
+      "loss": 0.8725,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986177058419001e-05,
+      "loss": 0.8806,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860735633362273e-05,
+      "loss": 0.9148,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859696849697718e-05,
+      "loss": 0.9371,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858654233600107e-05,
+      "loss": 0.9296,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857607785474713e-05,
+      "loss": 0.8616,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985655750572828e-05,
+      "loss": 0.8647,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985550339476905e-05,
+      "loss": 0.8968,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854445453006753e-05,
+      "loss": 0.8174,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985338368085261e-05,
+      "loss": 0.8796,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852318078719324e-05,
+      "loss": 0.9057,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985124864702109e-05,
+      "loss": 0.9206,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850175386173596e-05,
+      "loss": 0.8915,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984909829659402e-05,
+      "loss": 0.9303,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848017378701018e-05,
+      "loss": 0.8605,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846932632914733e-05,
+      "loss": 0.8456,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845844059656814e-05,
+      "loss": 0.9055,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984475165935038e-05,
+      "loss": 0.9258,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843655432420045e-05,
+      "loss": 0.9504,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842555379291907e-05,
+      "loss": 0.8699,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841451500393552e-05,
+      "loss": 0.9069,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984034379615406e-05,
+      "loss": 0.8771,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839232267003988e-05,
+      "loss": 0.858,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838116913375386e-05,
+      "loss": 0.7892,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836997735701788e-05,
+      "loss": 0.9075,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983587473441821e-05,
+      "loss": 0.8867,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834747909961166e-05,
+      "loss": 0.8942,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833617262768644e-05,
+      "loss": 0.8576,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983248279328013e-05,
+      "loss": 0.3191,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831344501936578e-05,
+      "loss": 0.9054,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983020238918045e-05,
+      "loss": 0.8402,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982905645545567e-05,
+      "loss": 0.952,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827906701207667e-05,
+      "loss": 0.925,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826753126883346e-05,
+      "loss": 0.8732,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825595732931092e-05,
+      "loss": 0.8265,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824434519800788e-05,
+      "loss": 0.8829,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823269487943787e-05,
+      "loss": 0.3123,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982210063781294e-05,
+      "loss": 0.8795,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820927969862573e-05,
+      "loss": 0.9486,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819751484548496e-05,
+      "loss": 0.8664,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818571182328003e-05,
+      "loss": 0.8869,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981738706365988e-05,
+      "loss": 0.9033,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816199129004387e-05,
+      "loss": 0.8652,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815007378823273e-05,
+      "loss": 0.8528,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981381181357976e-05,
+      "loss": 0.8475,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812612433738572e-05,
+      "loss": 0.8888,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811409239765896e-05,
+      "loss": 0.9024,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981020223212941e-05,
+      "loss": 0.9366,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808991411298276e-05,
+      "loss": 0.317,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807776777743136e-05,
+      "loss": 0.8908,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806558331936118e-05,
+      "loss": 0.9288,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805336074350817e-05,
+      "loss": 0.3161,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804110005462332e-05,
+      "loss": 0.8748,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802880125747227e-05,
+      "loss": 0.8815,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801646435683556e-05,
+      "loss": 0.8845,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800408935750846e-05,
+      "loss": 0.3007,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979916762643011e-05,
+      "loss": 0.3187,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797922508203843e-05,
+      "loss": 0.85,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979667358155602e-05,
+      "loss": 0.9135,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795420846972088e-05,
+      "loss": 0.8895,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794164304938985e-05,
+      "loss": 0.8874,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979290395594513e-05,
+      "loss": 0.8876,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791639800480408e-05,
+      "loss": 0.817,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97903718390362e-05,
+      "loss": 0.8484,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789100072105353e-05,
+      "loss": 0.867,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97878245001822e-05,
+      "loss": 0.946,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786545123762556e-05,
+      "loss": 0.8402,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785261943343702e-05,
+      "loss": 0.9234,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783974959424412e-05,
+      "loss": 0.8343,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782684172504933e-05,
+      "loss": 0.8575,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978138958308699e-05,
+      "loss": 0.9539,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780091191673784e-05,
+      "loss": 0.8401,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778788998769994e-05,
+      "loss": 0.3053,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777483004881785e-05,
+      "loss": 0.9223,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776173210516783e-05,
+      "loss": 0.8248,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774859616184106e-05,
+      "loss": 0.8729,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977354222239435e-05,
+      "loss": 0.9473,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977222102965957e-05,
+      "loss": 0.8938,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770896038493317e-05,
+      "loss": 0.8883,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976956724941061e-05,
+      "loss": 0.8931,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768234662927942e-05,
+      "loss": 0.9003,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976689827956329e-05,
+      "loss": 0.8774,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97655580998361e-05,
+      "loss": 0.9013,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976421412426729e-05,
+      "loss": 0.859,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976286635337927e-05,
+      "loss": 0.9093,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761514787695903e-05,
+      "loss": 0.8507,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760159427742547e-05,
+      "loss": 0.8671,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758800274046022e-05,
+      "loss": 0.8411,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757437327134627e-05,
+      "loss": 0.8418,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756070587538136e-05,
+      "loss": 0.9152,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754700055787798e-05,
+      "loss": 0.9432,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975332573241633e-05,
+      "loss": 0.9133,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975194761795793e-05,
+      "loss": 0.9162,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975056571294827e-05,
+      "loss": 0.8773,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749180017924484e-05,
+      "loss": 0.8153,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747790533425192e-05,
+      "loss": 0.9301,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746397259990485e-05,
+      "loss": 0.9095,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974500019816192e-05,
+      "loss": 0.8479,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974359934848253e-05,
+      "loss": 0.9315,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742194711496823e-05,
+      "loss": 0.8676,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740786287750778e-05,
+      "loss": 0.8798,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739374077791843e-05,
+      "loss": 0.782,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737958082168936e-05,
+      "loss": 0.887,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736538301432457e-05,
+      "loss": 0.29,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735114736134266e-05,
+      "loss": 0.9239,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97336873868277e-05,
+      "loss": 0.9339,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732256254067567e-05,
+      "loss": 0.9081,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973082133841014e-05,
+      "loss": 0.8444,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729382640413163e-05,
+      "loss": 0.9342,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727940160635862e-05,
+      "loss": 0.8586,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972649389963892e-05,
+      "loss": 0.8746,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725043857984493e-05,
+      "loss": 0.8792,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972359003623621e-05,
+      "loss": 0.9219,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722132434959163e-05,
+      "loss": 0.7583,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720671054719927e-05,
+      "loss": 0.9371,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719205896086522e-05,
+      "loss": 0.8875,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971773695962846e-05,
+      "loss": 0.9068,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971626424591671e-05,
+      "loss": 0.9541,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971478775552371e-05,
+      "loss": 0.8361,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713307489023368e-05,
+      "loss": 0.8068,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711823446991056e-05,
+      "loss": 0.9153,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710335630003623e-05,
+      "loss": 0.8693,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708844038639373e-05,
+      "loss": 0.8564,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707348673478083e-05,
+      "loss": 0.8386,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705849535101e-05,
+      "loss": 0.9306,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704346624090833e-05,
+      "loss": 0.8653,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702839941031758e-05,
+      "loss": 0.8931,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970132948650942e-05,
+      "loss": 0.8709,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699815261110925e-05,
+      "loss": 0.8811,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698297265424848e-05,
+      "loss": 0.8541,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696775500041227e-05,
+      "loss": 0.8098,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695249965551574e-05,
+      "loss": 0.8189,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969372066254886e-05,
+      "loss": 0.8849,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969218759162751e-05,
+      "loss": 0.8231,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690650753383434e-05,
+      "loss": 0.8537,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968911014841399e-05,
+      "loss": 0.8157,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687565777318008e-05,
+      "loss": 0.902,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686017640695783e-05,
+      "loss": 0.8775,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684465739149066e-05,
+      "loss": 0.8501,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968291007328108e-05,
+      "loss": 0.9344,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968135064369651e-05,
+      "loss": 0.9019,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679787451001492e-05,
+      "loss": 0.8369,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967822049580364e-05,
+      "loss": 0.8336,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676649778712023e-05,
+      "loss": 0.8854,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675075300337177e-05,
+      "loss": 0.8568,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673497061291096e-05,
+      "loss": 0.9353,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671915062187236e-05,
+      "loss": 0.8738,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670329303640508e-05,
+      "loss": 0.8514,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668739786267305e-05,
+      "loss": 0.8978,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667146510685457e-05,
+      "loss": 0.8936,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966554947751427e-05,
+      "loss": 0.829,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96639486873745e-05,
+      "loss": 0.3515,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662344140888373e-05,
+      "loss": 0.8761,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660735838679575e-05,
+      "loss": 0.8962,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659123781373243e-05,
+      "loss": 0.9082,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657507969595974e-05,
+      "loss": 0.8597,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965588840397584e-05,
+      "loss": 0.8796,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654265085142356e-05,
+      "loss": 0.9383,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96526380137265e-05,
+      "loss": 0.9321,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965100719036071e-05,
+      "loss": 0.8466,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649372615678883e-05,
+      "loss": 0.8901,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964773429031637e-05,
+      "loss": 0.8949,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646092214909985e-05,
+      "loss": 0.9166,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644446390098e-05,
+      "loss": 0.9311,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642796816520134e-05,
+      "loss": 0.3486,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964114349481758e-05,
+      "loss": 0.87,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639486425632978e-05,
+      "loss": 0.9392,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963782560961042e-05,
+      "loss": 0.2946,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636161047395464e-05,
+      "loss": 0.8562,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634492739635117e-05,
+      "loss": 0.8857,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963282068697785e-05,
+      "loss": 0.8512,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631144890073577e-05,
+      "loss": 0.859,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629465349573686e-05,
+      "loss": 0.8789,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627782066131e-05,
+      "loss": 0.8223,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626095040399807e-05,
+      "loss": 0.8809,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962440427303585e-05,
+      "loss": 0.8736,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622709764696323e-05,
+      "loss": 0.8654,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621011516039882e-05,
+      "loss": 0.8465,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619309527726624e-05,
+      "loss": 0.3519,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617603800418113e-05,
+      "loss": 0.8209,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961589433477735e-05,
+      "loss": 0.8238,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614181131468815e-05,
+      "loss": 0.9088,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961246419115841e-05,
+      "loss": 0.8672,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610743514513507e-05,
+      "loss": 0.8287,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960901910220293e-05,
+      "loss": 0.827,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607290954896952e-05,
+      "loss": 0.9177,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96055590732673e-05,
+      "loss": 0.8944,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603823457987153e-05,
+      "loss": 0.8934,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602084109731132e-05,
+      "loss": 0.8532,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600341029175323e-05,
+      "loss": 0.8783,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959859421699725e-05,
+      "loss": 0.8599,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95968436738759e-05,
+      "loss": 0.9189,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595089400491697e-05,
+      "loss": 0.9214,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593331397526524e-05,
+      "loss": 0.921,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959156966566371e-05,
+      "loss": 0.3353,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589804205588043e-05,
+      "loss": 0.854,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958803501798574e-05,
+      "loss": 0.8623,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586262103544484e-05,
+      "loss": 0.8032,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584485462953404e-05,
+      "loss": 0.3086,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582705096903068e-05,
+      "loss": 0.9015,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580921006085506e-05,
+      "loss": 0.8498,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957913319119418e-05,
+      "loss": 0.8589,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577341652924016e-05,
+      "loss": 0.9175,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575546391971376e-05,
+      "loss": 0.8658,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957374740903407e-05,
+      "loss": 0.8804,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957194470481136e-05,
+      "loss": 0.8059,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957013828000395e-05,
+      "loss": 0.8481,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568328135313995e-05,
+      "loss": 0.3546,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566514271445086e-05,
+      "loss": 0.8148,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564696689102274e-05,
+      "loss": 0.9339,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956287538899204e-05,
+      "loss": 0.9027,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561050371822325e-05,
+      "loss": 0.8523,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559221638302502e-05,
+      "loss": 0.8684,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557389189143392e-05,
+      "loss": 0.3436,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955555302505727e-05,
+      "loss": 0.8605,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955371314675784e-05,
+      "loss": 0.8471,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551869554960263e-05,
+      "loss": 0.8123,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955002225038113e-05,
+      "loss": 0.7927,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954817123373849e-05,
+      "loss": 0.9106,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546316505751823e-05,
+      "loss": 0.8448,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544458067142056e-05,
+      "loss": 0.8719,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954259591863156e-05,
+      "loss": 0.9116,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540730060944142e-05,
+      "loss": 0.8193,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953886049480506e-05,
+      "loss": 0.8531,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953698722094101e-05,
+      "loss": 0.9517,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953511024008012e-05,
+      "loss": 0.9112,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533229552951972e-05,
+      "loss": 0.8788,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531345160287582e-05,
+      "loss": 0.8554,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529457062819408e-05,
+      "loss": 0.9437,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527565261281346e-05,
+      "loss": 0.9279,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525669756408737e-05,
+      "loss": 0.8746,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523770548938352e-05,
+      "loss": 0.7763,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521867639608412e-05,
+      "loss": 0.8929,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951996102915857e-05,
+      "loss": 0.8122,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518050718329918e-05,
+      "loss": 0.8484,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951613670786499e-05,
+      "loss": 0.8952,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514218998507754e-05,
+      "loss": 0.8695,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951229759100362e-05,
+      "loss": 0.8785,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951037248609943e-05,
+      "loss": 0.8432,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508443684543468e-05,
+      "loss": 0.8725,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506511187085455e-05,
+      "loss": 0.8715,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950457499447654e-05,
+      "loss": 0.8446,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502635107469326e-05,
+      "loss": 0.7961,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950069152681783e-05,
+      "loss": 0.9227,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498744253277522e-05,
+      "loss": 0.8734,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496793287605294e-05,
+      "loss": 0.9102,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949483863055949e-05,
+      "loss": 0.8997,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492880282899876e-05,
+      "loss": 0.848,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949091824538765e-05,
+      "loss": 0.8925,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488952518785455e-05,
+      "loss": 0.8066,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948698310385736e-05,
+      "loss": 0.9014,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485010001368878e-05,
+      "loss": 0.8551,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483033212086936e-05,
+      "loss": 0.8829,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481052736779914e-05,
+      "loss": 0.8596,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947906857621761e-05,
+      "loss": 0.905,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477080731171267e-05,
+      "loss": 0.906,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947508920241355e-05,
+      "loss": 0.8953,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473093990718567e-05,
+      "loss": 0.8806,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471095096861843e-05,
+      "loss": 0.8384,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469092521620342e-05,
+      "loss": 0.7839,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946708626577246e-05,
+      "loss": 0.9181,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465076330098025e-05,
+      "loss": 0.9445,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946306271537829e-05,
+      "loss": 0.8756,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461045422395943e-05,
+      "loss": 0.904,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459024451935097e-05,
+      "loss": 0.8606,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456999804781298e-05,
+      "loss": 0.8675,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454971481721515e-05,
+      "loss": 0.3235,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945293948354416e-05,
+      "loss": 0.8609,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945090381103906e-05,
+      "loss": 0.8916,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448864464997474e-05,
+      "loss": 0.8695,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446821446212087e-05,
+      "loss": 0.8357,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944477475547702e-05,
+      "loss": 0.8683,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442724393587806e-05,
+      "loss": 0.934,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440670361341422e-05,
+      "loss": 0.8529,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438612659536265e-05,
+      "loss": 0.8941,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943655128897215e-05,
+      "loss": 0.8804,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943448625045033e-05,
+      "loss": 0.8649,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943241754477348e-05,
+      "loss": 0.8508,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430345172745694e-05,
+      "loss": 0.9434,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428269135172503e-05,
+      "loss": 0.9366,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426189432860853e-05,
+      "loss": 0.8552,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424106066619117e-05,
+      "loss": 0.9075,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422019037257093e-05,
+      "loss": 0.8719,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419928345586007e-05,
+      "loss": 0.8201,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417833992418495e-05,
+      "loss": 0.8954,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415735978568636e-05,
+      "loss": 0.925,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413634304851918e-05,
+      "loss": 0.9735,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411528972085248e-05,
+      "loss": 0.8473,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409419981086966e-05,
+      "loss": 0.283,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940730733267684e-05,
+      "loss": 0.868,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405191027676036e-05,
+      "loss": 0.823,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940307106690716e-05,
+      "loss": 0.8599,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400947451194242e-05,
+      "loss": 0.8167,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398820181362713e-05,
+      "loss": 0.8578,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939668925823944e-05,
+      "loss": 0.891,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394554682652705e-05,
+      "loss": 0.8878,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392416455432216e-05,
+      "loss": 0.9231,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939027457740909e-05,
+      "loss": 0.8992,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938812904941587e-05,
+      "loss": 0.8992,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385979872286516e-05,
+      "loss": 0.7955,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383827046856405e-05,
+      "loss": 0.8929,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381670573962338e-05,
+      "loss": 0.8263,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937951045444252e-05,
+      "loss": 0.8728,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937734668913659e-05,
+      "loss": 0.8577,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93751792788856e-05,
+      "loss": 0.8735,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937300822453201e-05,
+      "loss": 0.9075,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93708335269197e-05,
+      "loss": 0.904,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936865518689397e-05,
+      "loss": 0.9016,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366473205301536e-05,
+      "loss": 0.3002,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364287582990526e-05,
+      "loss": 0.8648,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362098320810486e-05,
+      "loss": 0.7829,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935990541961237e-05,
+      "loss": 0.8212,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357708880248563e-05,
+      "loss": 0.3666,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355508703572837e-05,
+      "loss": 0.8375,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935330489044041e-05,
+      "loss": 0.8845,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351097441707885e-05,
+      "loss": 0.8602,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348886358233295e-05,
+      "loss": 0.8445,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934667164087608e-05,
+      "loss": 0.7854,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344453290497097e-05,
+      "loss": 0.3554,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934223130795861e-05,
+      "loss": 0.7801,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340005694124294e-05,
+      "loss": 0.918,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337776449859243e-05,
+      "loss": 0.8432,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335543576029952e-05,
+      "loss": 0.8798,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333307073504336e-05,
+      "loss": 0.8399,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331066943151714e-05,
+      "loss": 0.8637,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328823185842816e-05,
+      "loss": 0.9074,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932657580244979e-05,
+      "loss": 0.8581,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932432479384618e-05,
+      "loss": 0.8459,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322070160906946e-05,
+      "loss": 0.909,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319811904508464e-05,
+      "loss": 0.8303,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317550025528505e-05,
+      "loss": 0.8506,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315284524846257e-05,
+      "loss": 0.7828,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931301540334231e-05,
+      "loss": 0.3199,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310742661898666e-05,
+      "loss": 0.8962,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308466301398734e-05,
+      "loss": 0.9208,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306186322727325e-05,
+      "loss": 0.863,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930390272677066e-05,
+      "loss": 0.7966,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930161551441637e-05,
+      "loss": 0.8244,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929932468655348e-05,
+      "loss": 0.8544,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297030244072436e-05,
+      "loss": 0.838,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929473218786507e-05,
+      "loss": 0.8769,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292430518824642e-05,
+      "loss": 0.8697,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290125237845793e-05,
+      "loss": 0.86,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928781634582458e-05,
+      "loss": 0.3383,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285503843658468e-05,
+      "loss": 0.8595,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283187732246315e-05,
+      "loss": 0.8725,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280868012488386e-05,
+      "loss": 0.9037,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278544685286347e-05,
+      "loss": 0.8035,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276217751543274e-05,
+      "loss": 0.8323,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927388721216363e-05,
+      "loss": 0.8931,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92715530680533e-05,
+      "loss": 0.9454,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269215320119547e-05,
+      "loss": 0.9164,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266873969271053e-05,
+      "loss": 0.8563,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926452901641789e-05,
+      "loss": 0.8457,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262180462471538e-05,
+      "loss": 0.8374,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925982830834487e-05,
+      "loss": 0.8878,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925747255495216e-05,
+      "loss": 0.8395,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255113203209083e-05,
+      "loss": 0.8959,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252750254032712e-05,
+      "loss": 0.9197,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925038370834152e-05,
+      "loss": 0.8976,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924801356705537e-05,
+      "loss": 0.8098,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924563983109553e-05,
+      "loss": 0.8493,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924326250138467e-05,
+      "loss": 0.7567,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240881578846838e-05,
+      "loss": 0.8181,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238497064407505e-05,
+      "loss": 0.8763,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236108958993517e-05,
+      "loss": 0.8627,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233717263533124e-05,
+      "loss": 0.8395,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923132197895597e-05,
+      "loss": 0.8565,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228923106193098e-05,
+      "loss": 0.8595,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226520646176937e-05,
+      "loss": 0.8623,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922411459984132e-05,
+      "loss": 0.8354,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221704968121467e-05,
+      "loss": 0.7692,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219291751953993e-05,
+      "loss": 0.9204,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216874952276914e-05,
+      "loss": 0.8835,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214454570029624e-05,
+      "loss": 0.8859,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212030606152922e-05,
+      "loss": 0.8337,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209603061588997e-05,
+      "loss": 0.8529,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207171937281423e-05,
+      "loss": 0.8666,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204737234175178e-05,
+      "loss": 0.8594,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202298953216613e-05,
+      "loss": 0.7946,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919985709535349e-05,
+      "loss": 0.8416,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197411661534945e-05,
+      "loss": 0.7889,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194962652711515e-05,
+      "loss": 0.8607,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919251006983512e-05,
+      "loss": 0.8246,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919005391385907e-05,
+      "loss": 0.8262,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918759418573806e-05,
+      "loss": 0.8425,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185130886428188e-05,
+      "loss": 0.8764,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918266401688693e-05,
+      "loss": 0.829,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180193578073146e-05,
+      "loss": 0.9154,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917771957094709e-05,
+      "loss": 0.8373,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91752419964704e-05,
+      "loss": 0.8309,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91727608556061e-05,
+      "loss": 0.8718,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170276149318606e-05,
+      "loss": 0.327,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916778787857371e-05,
+      "loss": 0.9145,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165296044338597e-05,
+      "loss": 0.8517,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162800647581837e-05,
+      "loss": 0.8451,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160301689273382e-05,
+      "loss": 0.8962,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915779917038457e-05,
+      "loss": 0.8528,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915529309188812e-05,
+      "loss": 0.8998,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152783454758133e-05,
+      "loss": 0.8633,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915027025997011e-05,
+      "loss": 0.9206,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147753508500907e-05,
+      "loss": 0.3498,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145233201328786e-05,
+      "loss": 0.8486,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914270933943338e-05,
+      "loss": 0.844,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140181923795706e-05,
+      "loss": 0.8472,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137650955398163e-05,
+      "loss": 0.8069,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135116435224535e-05,
+      "loss": 0.8937,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132578364259973e-05,
+      "loss": 0.331,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130036743491023e-05,
+      "loss": 0.8887,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127491573905603e-05,
+      "loss": 0.886,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124942856493013e-05,
+      "loss": 0.8752,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912239059224393e-05,
+      "loss": 0.8683,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119834782150416e-05,
+      "loss": 0.3241,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117275427205904e-05,
+      "loss": 0.8625,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114712528405202e-05,
+      "loss": 0.8918,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112146086744504e-05,
+      "loss": 0.8895,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910957610322138e-05,
+      "loss": 0.9532,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107002578834778e-05,
+      "loss": 0.3134,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104425514585005e-05,
+      "loss": 0.8167,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910184491147377e-05,
+      "loss": 0.8157,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099260770504143e-05,
+      "loss": 0.7721,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096673092680572e-05,
+      "loss": 0.8635,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094081879008873e-05,
+      "loss": 0.9097,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909148713049625e-05,
+      "loss": 0.8569,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908888884815127e-05,
+      "loss": 0.9065,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086287032983878e-05,
+      "loss": 0.8263,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908368168600539e-05,
+      "loss": 0.8633,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081072808228497e-05,
+      "loss": 0.8645,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078460400667263e-05,
+      "loss": 0.8403,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075844464337117e-05,
+      "loss": 0.8741,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073225000254876e-05,
+      "loss": 0.9036,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070602009438707e-05,
+      "loss": 0.8463,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067975492908167e-05,
+      "loss": 0.8335,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065345451684166e-05,
+      "loss": 0.9443,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062711886788997e-05,
+      "loss": 0.8849,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060074799246316e-05,
+      "loss": 0.8352,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057434190081154e-05,
+      "loss": 0.8014,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054790060319908e-05,
+      "loss": 0.9129,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905214241099034e-05,
+      "loss": 0.8622,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904949124312158e-05,
+      "loss": 0.8384,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046836557744135e-05,
+      "loss": 0.854,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904417835588987e-05,
+      "loss": 0.8863,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041516638592025e-05,
+      "loss": 0.8623,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903885140688519e-05,
+      "loss": 0.8341,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903618266180534e-05,
+      "loss": 0.8457,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033510404389806e-05,
+      "loss": 0.8913,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030834635677285e-05,
+      "loss": 0.8981,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028155356707843e-05,
+      "loss": 0.8526,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025472568522906e-05,
+      "loss": 0.9078,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902278627216526e-05,
+      "loss": 0.8653,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902009646867907e-05,
+      "loss": 0.8539,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017403159109844e-05,
+      "loss": 0.8612,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014706344504467e-05,
+      "loss": 0.8734,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901200602591119e-05,
+      "loss": 0.8825,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009302204379604e-05,
+      "loss": 0.863,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900659488096068e-05,
+      "loss": 0.8786,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003884056706758e-05,
+      "loss": 0.8645,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900116973267151e-05,
+      "loss": 0.8719,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899845190991e-05,
+      "loss": 0.8176,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995730589478628e-05,
+      "loss": 0.918,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993005772435162e-05,
+      "loss": 0.8814,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990277459838734e-05,
+      "loss": 0.9081,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898754565274983e-05,
+      "loss": 0.8589,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984810352230295e-05,
+      "loss": 0.7933,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898207155934333e-05,
+      "loss": 0.8472,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979329275153492e-05,
+      "loss": 0.8595,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976583500726706e-05,
+      "loss": 0.8682,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897383423713024e-05,
+      "loss": 0.299,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897108148543273e-05,
+      "loss": 0.9291,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896832524670415e-05,
+      "loss": 0.8688,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965565522015856e-05,
+      "loss": 0.8552,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962802312440533e-05,
+      "loss": 0.8913,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896003561905224e-05,
+      "loss": 0.8908,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957265442926376e-05,
+      "loss": 0.885,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954491785139703e-05,
+      "loss": 0.9008,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951714646770332e-05,
+      "loss": 0.3521,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948934028897725e-05,
+      "loss": 0.8099,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946149932602708e-05,
+      "loss": 0.8472,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943362358967442e-05,
+      "loss": 0.887,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940571309075445e-05,
+      "loss": 0.3621,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937776784011603e-05,
+      "loss": 0.9009,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893497878486213e-05,
+      "loss": 0.9433,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932177312714595e-05,
+      "loss": 0.8978,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929372368657933e-05,
+      "loss": 0.8311,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926563953782407e-05,
+      "loss": 0.7797,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923752069179646e-05,
+      "loss": 0.8352,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920936715942614e-05,
+      "loss": 0.8925,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918117895165636e-05,
+      "loss": 0.3572,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915295607944374e-05,
+      "loss": 0.8952,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912469855375847e-05,
+      "loss": 0.8691,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909640638558415e-05,
+      "loss": 0.8819,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890680795859178e-05,
+      "loss": 0.9105,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903971816577e-05,
+      "loss": 0.7941,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901132213616477e-05,
+      "loss": 0.8414,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898289150813954e-05,
+      "loss": 0.9144,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895442629274516e-05,
+      "loss": 0.8588,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88925926501046e-05,
+      "loss": 0.8372,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889739214411987e-05,
+      "loss": 0.8374,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886882323305796e-05,
+      "loss": 0.8889,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884021977896493e-05,
+      "loss": 0.8558,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888115817929588e-05,
+      "loss": 0.8662,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887829092861711e-05,
+      "loss": 0.919,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875420226974676e-05,
+      "loss": 0.8383,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887254607548441e-05,
+      "loss": 0.844,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869668475263484e-05,
+      "loss": 0.3199,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886678742743041e-05,
+      "loss": 0.9246,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886390293310505e-05,
+      "loss": 0.8539,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886101499340859e-05,
+      "loss": 0.8671,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858123609463565e-05,
+      "loss": 0.8853,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885522878239385e-05,
+      "loss": 0.7837,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885233051332465e-05,
+      "loss": 0.3157,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884942880338252e-05,
+      "loss": 0.9363,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884652365369534e-05,
+      "loss": 0.8768,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884361506539233e-05,
+      "loss": 0.9058,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840703039604058e-05,
+      "loss": 0.8706,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883778757746241e-05,
+      "loss": 0.8331,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834868680100623e-05,
+      "loss": 0.8923,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831946348653263e-05,
+      "loss": 0.8472,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882902058425623e-05,
+      "loss": 0.8783,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826091388046753e-05,
+      "loss": 0.9299,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823158761163415e-05,
+      "loss": 0.8939,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820222704746106e-05,
+      "loss": 0.8238,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817283219936066e-05,
+      "loss": 0.8197,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814340307875866e-05,
+      "loss": 0.8858,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811393969709405e-05,
+      "loss": 0.8629,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808444206581918e-05,
+      "loss": 0.8393,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880549101963996e-05,
+      "loss": 0.834,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802534410031436e-05,
+      "loss": 0.8139,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799574378905567e-05,
+      "loss": 0.9209,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87966109274129e-05,
+      "loss": 0.3579,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879364405670533e-05,
+      "loss": 0.9085,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790673767936068e-05,
+      "loss": 0.8632,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787700062259652e-05,
+      "loss": 0.9058,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784722940831956e-05,
+      "loss": 0.8989,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781742404810176e-05,
+      "loss": 0.8605,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778758455352836e-05,
+      "loss": 0.8035,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877577109361979e-05,
+      "loss": 0.8428,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772780320772214e-05,
+      "loss": 0.7694,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876978613797261e-05,
+      "loss": 0.8712,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8766788546384815e-05,
+      "loss": 0.7387,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763787547173976e-05,
+      "loss": 0.8489,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760783141506573e-05,
+      "loss": 0.876,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757775330550413e-05,
+      "loss": 0.8274,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754764115474616e-05,
+      "loss": 0.8405,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751749497449642e-05,
+      "loss": 0.8775,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748731477647253e-05,
+      "loss": 0.7379,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874571005724055e-05,
+      "loss": 0.78,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742685237403946e-05,
+      "loss": 0.8452,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739657019313183e-05,
+      "loss": 0.8541,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736625404145316e-05,
+      "loss": 0.7895,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733590393078723e-05,
+      "loss": 0.7962,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730551987293112e-05,
+      "loss": 0.8078,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872751018796949e-05,
+      "loss": 0.862,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724464996290203e-05,
+      "loss": 0.8537,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721416413438906e-05,
+      "loss": 0.3312,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871836444060057e-05,
+      "loss": 0.866,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715309078961494e-05,
+      "loss": 0.8799,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871225032970928e-05,
+      "loss": 0.84,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709188194032854e-05,
+      "loss": 0.8685,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706122673122464e-05,
+      "loss": 0.8335,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870305376816967e-05,
+      "loss": 0.912,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699981480367335e-05,
+      "loss": 0.8196,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869690581090966e-05,
+      "loss": 0.8525,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869382676099214e-05,
+      "loss": 0.858,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690744331811592e-05,
+      "loss": 0.8864,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868765852456615e-05,
+      "loss": 0.3553,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684569340455256e-05,
+      "loss": 0.8262,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681476780679668e-05,
+      "loss": 0.8474,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678380846441457e-05,
+      "loss": 0.8359,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867528153894399e-05,
+      "loss": 0.8274,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672178859391968e-05,
+      "loss": 0.8167,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669072808991394e-05,
+      "loss": 0.8257,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665963388949577e-05,
+      "loss": 0.9409,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662850600475138e-05,
+      "loss": 0.7781,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659734444778014e-05,
+      "loss": 0.7637,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656614923069433e-05,
+      "loss": 0.7863,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653492036561952e-05,
+      "loss": 0.8616,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650365786469425e-05,
+      "loss": 0.8259,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864723617400702e-05,
+      "loss": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86441032003912e-05,
+      "loss": 0.8851,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640966866839746e-05,
+      "loss": 0.8573,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863782717457174e-05,
+      "loss": 0.863,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863468412480757e-05,
+      "loss": 0.9101,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631537718768927e-05,
+      "loss": 0.8382,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862838795767882e-05,
+      "loss": 0.8452,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862523484276154e-05,
+      "loss": 0.9303,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622078375242696e-05,
+      "loss": 0.809,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86189185563492e-05,
+      "loss": 0.3381,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615755387309257e-05,
+      "loss": 0.8538,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612588869352387e-05,
+      "loss": 0.8557,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86094190037094e-05,
+      "loss": 0.8144,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860624579161242e-05,
+      "loss": 0.8194,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860306923429486e-05,
+      "loss": 0.8119,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599889332991438e-05,
+      "loss": 0.8443,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596706088938172e-05,
+      "loss": 0.8785,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593519503372378e-05,
+      "loss": 0.88,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590329577532674e-05,
+      "loss": 0.3377,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587136312658972e-05,
+      "loss": 0.913,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583939709992486e-05,
+      "loss": 0.8489,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580739770775723e-05,
+      "loss": 0.8194,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577536496252492e-05,
+      "loss": 0.8467,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857432988766789e-05,
+      "loss": 0.8105,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571119946268327e-05,
+      "loss": 0.7937,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567906673301484e-05,
+      "loss": 0.7916,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564690070016358e-05,
+      "loss": 0.8388,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856147013766323e-05,
+      "loss": 0.8778,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855824687749368e-05,
+      "loss": 0.8484,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555020290760572e-05,
+      "loss": 0.8216,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855179037871808e-05,
+      "loss": 0.8984,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548557142621656e-05,
+      "loss": 0.3416,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545320583728044e-05,
+      "loss": 0.8663,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542080703295293e-05,
+      "loss": 0.7901,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538837502582727e-05,
+      "loss": 0.8488,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535590982850972e-05,
+      "loss": 0.888,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853234114536194e-05,
+      "loss": 0.8671,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852908799137883e-05,
+      "loss": 0.849,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525831522166135e-05,
+      "loss": 0.8272,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852257173898963e-05,
+      "loss": 0.9015,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851930864311639e-05,
+      "loss": 0.8536,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516042235814767e-05,
+      "loss": 0.9204,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85127725183544e-05,
+      "loss": 0.8572,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509499492006216e-05,
+      "loss": 0.8781,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506223158042442e-05,
+      "loss": 0.3627,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502943517736566e-05,
+      "loss": 0.8415,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499660572363377e-05,
+      "loss": 0.8513,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849637432319895e-05,
+      "loss": 0.8281,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493084771520635e-05,
+      "loss": 0.787,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489791918607075e-05,
+      "loss": 0.8062,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486495765738187e-05,
+      "loss": 0.8704,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483196314195174e-05,
+      "loss": 0.8982,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479893565260525e-05,
+      "loss": 0.8863,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476587520218013e-05,
+      "loss": 0.8176,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847327818035268e-05,
+      "loss": 0.8758,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846996554695086e-05,
+      "loss": 0.7509,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846664962130016e-05,
+      "loss": 0.8739,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846333040468947e-05,
+      "loss": 0.9353,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846000789840896e-05,
+      "loss": 0.7709,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845668210375008e-05,
+      "loss": 0.9327,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453353022005557e-05,
+      "loss": 0.7893,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450020654469388e-05,
+      "loss": 0.8548,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844668500243686e-05,
+      "loss": 0.9116,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443346067204525e-05,
+      "loss": 0.8493,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440003850070222e-05,
+      "loss": 0.8639,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8436658352333056e-05,
+      "loss": 0.8625,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433309575293414e-05,
+      "loss": 0.8697,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429957520252953e-05,
+      "loss": 0.8679,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842660218851461e-05,
+      "loss": 0.8052,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842324358138259e-05,
+      "loss": 0.8197,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419881700162365e-05,
+      "loss": 0.7642,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84165165461607e-05,
+      "loss": 0.9011,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413148120685616e-05,
+      "loss": 0.8259,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409776425046404e-05,
+      "loss": 0.8291,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840640146055364e-05,
+      "loss": 0.3419,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403023228519155e-05,
+      "loss": 0.8664,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399641730256057e-05,
+      "loss": 0.7887,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839625696707873e-05,
+      "loss": 0.8412,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392868940302815e-05,
+      "loss": 0.8666,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838947765124523e-05,
+      "loss": 0.8139,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386083101224158e-05,
+      "loss": 0.8499,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838268529155905e-05,
+      "loss": 0.9054,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379284223570624e-05,
+      "loss": 0.8872,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375879898580862e-05,
+      "loss": 0.8167,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372472317913018e-05,
+      "loss": 0.861,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369061482891606e-05,
+      "loss": 0.8831,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365647394842404e-05,
+      "loss": 0.8453,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362230055092463e-05,
+      "loss": 0.8655,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835880946497009e-05,
+      "loss": 0.8403,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355385625804854e-05,
+      "loss": 0.8686,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351958538927598e-05,
+      "loss": 0.8382,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834852820567041e-05,
+      "loss": 0.3619,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345094627366657e-05,
+      "loss": 0.8326,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341657805350957e-05,
+      "loss": 0.9239,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338217740959197e-05,
+      "loss": 0.8547,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334774435528514e-05,
+      "loss": 0.854,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833132789039731e-05,
+      "loss": 0.8753,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327878106905242e-05,
+      "loss": 0.903,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324425086393243e-05,
+      "loss": 0.8963,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320968830203482e-05,
+      "loss": 0.8599,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317509339679392e-05,
+      "loss": 0.8152,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314046616165675e-05,
+      "loss": 0.8169,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310580661008272e-05,
+      "loss": 0.3429,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307111475554396e-05,
+      "loss": 0.804,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83036390611525e-05,
+      "loss": 0.8357,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300163419152316e-05,
+      "loss": 0.7381,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296684550904794e-05,
+      "loss": 0.8823,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293202457762177e-05,
+      "loss": 0.8476,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828971714107793e-05,
+      "loss": 0.864,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82862286022068e-05,
+      "loss": 0.869,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828273684250476e-05,
+      "loss": 0.3332,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279241863329043e-05,
+      "loss": 0.8251,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275743666038144e-05,
+      "loss": 0.8337,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82722422519918e-05,
+      "loss": 0.8306,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268737622550996e-05,
+      "loss": 0.8946,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265229779077976e-05,
+      "loss": 0.8766,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261718722936225e-05,
+      "loss": 0.8557,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258204455490475e-05,
+      "loss": 0.7926,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254686978106717e-05,
+      "loss": 0.8011,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825116629215218e-05,
+      "loss": 0.8444,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247642398995347e-05,
+      "loss": 0.798,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824411530000594e-05,
+      "loss": 0.8479,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240584996554932e-05,
+      "loss": 0.8252,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237051490014545e-05,
+      "loss": 0.8526,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233514781758233e-05,
+      "loss": 0.7159,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822997487316071e-05,
+      "loss": 0.8355,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822643176559793e-05,
+      "loss": 0.8689,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222885460447083e-05,
+      "loss": 0.8721,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219335959086606e-05,
+      "loss": 0.8532,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821578326289618e-05,
+      "loss": 0.8934,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821222737325673e-05,
+      "loss": 0.8853,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208668291550414e-05,
+      "loss": 0.882,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820510601916064e-05,
+      "loss": 0.8607,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820154055747205e-05,
+      "loss": 0.8477,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197971907870528e-05,
+      "loss": 0.8457,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194400071743196e-05,
+      "loss": 0.8153,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190825050478414e-05,
+      "loss": 0.8487,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187246845465787e-05,
+      "loss": 0.8741,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183665458096147e-05,
+      "loss": 0.8867,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180080889761573e-05,
+      "loss": 0.8081,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176493141855368e-05,
+      "loss": 0.8538,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172902215772084e-05,
+      "loss": 0.886,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81693081129075e-05,
+      "loss": 0.8635,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165710834658635e-05,
+      "loss": 0.7658,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162110382423738e-05,
+      "loss": 0.8474,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158506757602296e-05,
+      "loss": 0.8658,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815489996159502e-05,
+      "loss": 0.3622,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151289995803866e-05,
+      "loss": 0.3502,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147676861632014e-05,
+      "loss": 0.3459,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144060560483878e-05,
+      "loss": 0.8128,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140441093765103e-05,
+      "loss": 0.8544,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813681846288257e-05,
+      "loss": 0.8683,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133192669244373e-05,
+      "loss": 0.8624,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129563714259854e-05,
+      "loss": 0.8357,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125931599339575e-05,
+      "loss": 0.8066,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122296325895327e-05,
+      "loss": 0.8097,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811865789534013e-05,
+      "loss": 0.8412,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811501630908823e-05,
+      "loss": 0.8806,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111371568555102e-05,
+      "loss": 0.8731,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107723675157442e-05,
+      "loss": 0.8562,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104072630313176e-05,
+      "loss": 0.8821,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100418435441454e-05,
+      "loss": 0.7924,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809676109196265e-05,
+      "loss": 0.8089,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809310060129836e-05,
+      "loss": 0.8847,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089436964871407e-05,
+      "loss": 0.7599,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085770184105837e-05,
+      "loss": 0.8259,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808210026042691e-05,
+      "loss": 0.7809,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807842719526112e-05,
+      "loss": 0.3371,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074750990036176e-05,
+      "loss": 0.7735,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8071071646181e-05,
+      "loss": 0.9058,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067389165125752e-05,
+      "loss": 0.8269,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063703548301787e-05,
+      "loss": 0.8383,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060014797141705e-05,
+      "loss": 0.859,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056322913079303e-05,
+      "loss": 0.8495,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052627897549614e-05,
+      "loss": 0.8975,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048929751988873e-05,
+      "loss": 0.9081,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045228477834536e-05,
+      "loss": 0.7825,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041524076525285e-05,
+      "loss": 0.8649,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037816549500996e-05,
+      "loss": 0.8454,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034105898202785e-05,
+      "loss": 0.8651,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030392124072963e-05,
+      "loss": 0.8356,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026675228555066e-05,
+      "loss": 0.807,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802295521309384e-05,
+      "loss": 0.8243,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801923207913524e-05,
+      "loss": 0.8374,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015505828126438e-05,
+      "loss": 0.8861,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801177646151582e-05,
+      "loss": 0.863,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008043980752967e-05,
+      "loss": 0.816,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004308387288697e-05,
+      "loss": 0.8032,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000569682575016e-05,
+      "loss": 0.894,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996827868065145e-05,
+      "loss": 0.8213,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993082945213522e-05,
+      "loss": 0.9078,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798933491547578e-05,
+      "loss": 0.8524,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798558378030877e-05,
+      "loss": 0.813,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981829541170547e-05,
+      "loss": 0.8183,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797807219952037e-05,
+      "loss": 0.8915,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797431175681871e-05,
+      "loss": 0.7851,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797054821452723e-05,
+      "loss": 0.7419,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966781574108815e-05,
+      "loss": 0.8227,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963011837027544e-05,
+      "loss": 0.8853,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79592390047487e-05,
+      "loss": 0.8425,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955463078738777e-05,
+      "loss": 0.7913,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951684060465456e-05,
+      "loss": 0.9022,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794790195139764e-05,
+      "loss": 0.8093,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944116753005416e-05,
+      "loss": 0.7606,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794032846676008e-05,
+      "loss": 0.8552,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936537094134127e-05,
+      "loss": 0.8138,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932742636601252e-05,
+      "loss": 0.8856,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792894509563635e-05,
+      "loss": 0.8545,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792514447271551e-05,
+      "loss": 0.8131,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921340769316024e-05,
+      "loss": 0.804,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791753398691638e-05,
+      "loss": 0.8498,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913724126996263e-05,
+      "loss": 0.7963,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909911191036548e-05,
+      "loss": 0.8381,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906095180519317e-05,
+      "loss": 0.8528,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790227609692784e-05,
+      "loss": 0.9546,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898453941746585e-05,
+      "loss": 0.8656,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894628716461205e-05,
+      "loss": 0.8456,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890800422558562e-05,
+      "loss": 0.8299,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788696906152669e-05,
+      "loss": 0.8491,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788313463485484e-05,
+      "loss": 0.7875,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787929714403343e-05,
+      "loss": 0.8637,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787545659055409e-05,
+      "loss": 0.8407,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871612975909623e-05,
+      "loss": 0.9061,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786776630159404e-05,
+      "loss": 0.7992,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786391656910252e-05,
+      "loss": 0.8352,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786006377993145e-05,
+      "loss": 0.8506,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785620793557839e-05,
+      "loss": 0.8128,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78523490375421e-05,
+      "loss": 0.7713,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784848708732252e-05,
+      "loss": 0.8171,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844622086420775e-05,
+      "loss": 0.8669,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840754036339185e-05,
+      "loss": 0.8344,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783688293858124e-05,
+      "loss": 0.8745,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833008794651632e-05,
+      "loss": 0.8165,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829131606056228e-05,
+      "loss": 0.8336,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825251374302073e-05,
+      "loss": 0.8216,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821368100897404e-05,
+      "loss": 0.8257,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781748178735164e-05,
+      "loss": 0.8546,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781359243517537e-05,
+      "loss": 0.3586,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809700045880383e-05,
+      "loss": 0.7748,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805804620979633e-05,
+      "loss": 0.8232,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780190616198726e-05,
+      "loss": 0.8706,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798004670418587e-05,
+      "loss": 0.8,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794100147790105e-05,
+      "loss": 0.8095,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790192595619494e-05,
+      "loss": 0.8747,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786282015425605e-05,
+      "loss": 0.8768,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782368408728468e-05,
+      "loss": 0.3282,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778451777049288e-05,
+      "loss": 0.7552,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774532121910452e-05,
+      "loss": 0.8705,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770609444835517e-05,
+      "loss": 0.8292,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776668374734921e-05,
+      "loss": 0.8352,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762755030977444e-05,
+      "loss": 0.3442,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758823297247293e-05,
+      "loss": 0.3673,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754888547687013e-05,
+      "loss": 0.8455,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750950783826027e-05,
+      "loss": 0.8244,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747010007194934e-05,
+      "loss": 0.8982,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743066219325497e-05,
+      "loss": 0.8499,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739119421750657e-05,
+      "loss": 0.8174,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735169616004522e-05,
+      "loss": 0.8735,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731216803622367e-05,
+      "loss": 0.3373,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727260986140642e-05,
+      "loss": 0.8515,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723302165096953e-05,
+      "loss": 0.8544,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771934034203009e-05,
+      "loss": 0.8874,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715375518479995e-05,
+      "loss": 0.8164,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711407695987783e-05,
+      "loss": 0.8259,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707436876095736e-05,
+      "loss": 0.8008,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77034630603473e-05,
+      "loss": 0.84,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699486250287083e-05,
+      "loss": 0.9448,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769550644746086e-05,
+      "loss": 0.8564,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691523653415562e-05,
+      "loss": 0.87,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687537869699296e-05,
+      "loss": 0.8359,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768354909786132e-05,
+      "loss": 0.8173,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679557339452057e-05,
+      "loss": 0.8055,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675562596023095e-05,
+      "loss": 0.8333,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767156486912717e-05,
+      "loss": 0.8015,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667564160318195e-05,
+      "loss": 0.9035,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663560471151227e-05,
+      "loss": 0.8209,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659553803182486e-05,
+      "loss": 0.8629,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655544157969355e-05,
+      "loss": 0.8952,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765153153707037e-05,
+      "loss": 0.8408,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647515942045227e-05,
+      "loss": 0.7322,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643497374454767e-05,
+      "loss": 0.8446,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639475835861e-05,
+      "loss": 0.3531,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635451327827087e-05,
+      "loss": 0.8548,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631423851917334e-05,
+      "loss": 0.8773,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627393409697217e-05,
+      "loss": 0.8158,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762336000273335e-05,
+      "loss": 0.9112,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761932363259351e-05,
+      "loss": 0.3383,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761528430084662e-05,
+      "loss": 0.7987,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611242009062754e-05,
+      "loss": 0.8668,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760719675881314e-05,
+      "loss": 0.8256,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603148551670152e-05,
+      "loss": 0.8235,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599097389207313e-05,
+      "loss": 0.3011,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595043272999306e-05,
+      "loss": 0.7943,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590986204621943e-05,
+      "loss": 0.8306,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75869261856522e-05,
+      "loss": 0.8483,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582863217668192e-05,
+      "loss": 0.7897,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757879730224918e-05,
+      "loss": 0.8341,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574728440975574e-05,
+      "loss": 0.7875,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757065663542893e-05,
+      "loss": 0.7519,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756658188719194e-05,
+      "loss": 0.848,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562504197848452e-05,
+      "loss": 0.8739,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558423568983446e-05,
+      "loss": 0.8052,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7554340002183053e-05,
+      "loss": 0.8306,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755025349903454e-05,
+      "loss": 0.8722,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546164061126322e-05,
+      "loss": 0.86,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542071690047947e-05,
+      "loss": 0.8271,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537976387390107e-05,
+      "loss": 0.7925,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533878154744634e-05,
+      "loss": 0.8956,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75297769937045e-05,
+      "loss": 0.9157,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525672905863808e-05,
+      "loss": 0.8371,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521565892817808e-05,
+      "loss": 0.8937,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751745595616288e-05,
+      "loss": 0.8619,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751334309749655e-05,
+      "loss": 0.8219,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509227318417465e-05,
+      "loss": 0.9309,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505108620525425e-05,
+      "loss": 0.8835,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500987005421345e-05,
+      "loss": 0.7528,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496862474707287e-05,
+      "loss": 0.9022,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492735029986442e-05,
+      "loss": 0.8602,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488604672863144e-05,
+      "loss": 0.8035,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484471404942836e-05,
+      "loss": 0.8347,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480335227832115e-05,
+      "loss": 0.3505,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476196143138695e-05,
+      "loss": 0.8226,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472054152471432e-05,
+      "loss": 0.8223,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74679092574403e-05,
+      "loss": 0.7896,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463761459656402e-05,
+      "loss": 0.7952,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459610760731984e-05,
+      "loss": 0.8947,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74554571622804e-05,
+      "loss": 0.886,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745130066591615e-05,
+      "loss": 0.8528,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447141273254845e-05,
+      "loss": 0.8564,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442978985913227e-05,
+      "loss": 0.7887,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7438813805509168e-05,
+      "loss": 0.8698,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743464573366166e-05,
+      "loss": 0.8611,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430474771990816e-05,
+      "loss": 0.8163,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426300922117877e-05,
+      "loss": 0.3127,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422124185665203e-05,
+      "loss": 0.8206,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417944564256285e-05,
+      "loss": 0.8603,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741376205951572e-05,
+      "loss": 0.8452,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409576673069247e-05,
+      "loss": 0.7905,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405388406543697e-05,
+      "loss": 0.836,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401197261567048e-05,
+      "loss": 0.8638,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739700323976838e-05,
+      "loss": 0.8769,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392806342777897e-05,
+      "loss": 0.8055,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388606572226912e-05,
+      "loss": 0.8202,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384403929747877e-05,
+      "loss": 0.8687,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380198416974338e-05,
+      "loss": 0.7558,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737599003554096e-05,
+      "loss": 0.8559,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371778787083538e-05,
+      "loss": 0.8379,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367564673238962e-05,
+      "loss": 0.8079,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736334769564525e-05,
+      "loss": 0.8455,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359127855941523e-05,
+      "loss": 0.8502,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354905155768027e-05,
+      "loss": 0.8401,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350679596766108e-05,
+      "loss": 0.8448,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346451180578222e-05,
+      "loss": 0.8389,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734221990884795e-05,
+      "loss": 0.8104,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733798578321997e-05,
+      "loss": 0.7724,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333748805340072e-05,
+      "loss": 0.8288,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329508976855162e-05,
+      "loss": 0.8444,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325266299413237e-05,
+      "loss": 0.8089,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732102077466342e-05,
+      "loss": 0.9372,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316772404255934e-05,
+      "loss": 0.7997,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312521189842108e-05,
+      "loss": 0.8441,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730826713307437e-05,
+      "loss": 0.8962,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304010235606256e-05,
+      "loss": 0.342,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729975049909242e-05,
+      "loss": 0.3513,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295487925188598e-05,
+      "loss": 0.8212,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291222515551643e-05,
+      "loss": 0.8856,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286954271839507e-05,
+      "loss": 0.8355,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728268319571124e-05,
+      "loss": 0.8007,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278409288827002e-05,
+      "loss": 0.8612,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274132552848042e-05,
+      "loss": 0.8217,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7269852989436713e-05,
+      "loss": 0.7985,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726557060025647e-05,
+      "loss": 0.776,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7261285386971863e-05,
+      "loss": 0.8674,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725699735124854e-05,
+      "loss": 0.8561,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252706494753254e-05,
+      "loss": 0.8546,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7248412819153835e-05,
+      "loss": 0.8198,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724411632611923e-05,
+      "loss": 0.8473,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239817017319473e-05,
+      "loss": 0.3701,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235514894425687e-05,
+      "loss": 0.778,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231209959110097e-05,
+      "loss": 0.7971,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226902213046016e-05,
+      "loss": 0.8318,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722259165790785e-05,
+      "loss": 0.7672,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72182782953711e-05,
+      "loss": 0.8426,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213962127112355e-05,
+      "loss": 0.8801,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209643154809298e-05,
+      "loss": 0.8808,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72053213801407e-05,
+      "loss": 0.8706,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200996804786423e-05,
+      "loss": 0.9261,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719666943042741e-05,
+      "loss": 0.8415,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71923392587457e-05,
+      "loss": 0.8414,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718800629142442e-05,
+      "loss": 0.3586,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718367053014778e-05,
+      "loss": 0.8459,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717933197660107e-05,
+      "loss": 0.8435,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174990632470682e-05,
+      "loss": 0.8812,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717064649944408e-05,
+      "loss": 0.8035,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166299579209806e-05,
+      "loss": 0.828,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716194987345751e-05,
+      "loss": 0.8087,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715759738387789e-05,
+      "loss": 0.8639,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153242112162763e-05,
+      "loss": 0.8402,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148884060004997e-05,
+      "loss": 0.8295,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714452322909856e-05,
+      "loss": 0.9188,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714015962113849e-05,
+      "loss": 0.8556,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713579323782091e-05,
+      "loss": 0.8889,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131424080843016e-05,
+      "loss": 0.799,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127052151903087e-05,
+      "loss": 0.7592,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122677452700477e-05,
+      "loss": 0.8025,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118299984935622e-05,
+      "loss": 0.8254,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113919750310026e-05,
+      "loss": 0.8036,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109536750526273e-05,
+      "loss": 0.861,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710515098728802e-05,
+      "loss": 0.7719,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100762462300005e-05,
+      "loss": 0.7513,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096371177268026e-05,
+      "loss": 0.8831,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091977133898967e-05,
+      "loss": 0.8738,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087580333900778e-05,
+      "loss": 0.7846,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083180778982477e-05,
+      "loss": 0.7587,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078778470854163e-05,
+      "loss": 0.7934,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074373411226998e-05,
+      "loss": 0.8597,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706996560181321e-05,
+      "loss": 0.848,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706555504432611e-05,
+      "loss": 0.8566,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706114174048006e-05,
+      "loss": 0.8818,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056725691990496e-05,
+      "loss": 0.8802,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052306900573926e-05,
+      "loss": 0.8055,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704788536794792e-05,
+      "loss": 0.8171,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043461095831113e-05,
+      "loss": 0.8526,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039034085943206e-05,
+      "loss": 0.8427,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034604340004967e-05,
+      "loss": 0.7494,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030171859738217e-05,
+      "loss": 0.8387,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025736646865852e-05,
+      "loss": 0.8282,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021298703111824e-05,
+      "loss": 0.8765,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7016858030201154e-05,
+      "loss": 0.7426,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012414629859905e-05,
+      "loss": 0.86,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700796850381522e-05,
+      "loss": 0.8766,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.70035196537953e-05,
+      "loss": 0.8494,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999068081529385e-05,
+      "loss": 0.7478,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994613788747797e-05,
+      "loss": 0.9024,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990156777181903e-05,
+      "loss": 0.8399,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698569704856413e-05,
+      "loss": 0.3553,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698123460462796e-05,
+      "loss": 0.847,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697676944710793e-05,
+      "loss": 0.8967,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697230157773963e-05,
+      "loss": 0.3529,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967830998259715e-05,
+      "loss": 0.8653,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696335771040588e-05,
+      "loss": 0.3679,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958881715916878e-05,
+      "loss": 0.3245,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695440301653251e-05,
+      "loss": 0.8619,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949921613993644e-05,
+      "loss": 0.7861,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694543751004217e-05,
+      "loss": 0.8238,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940950706421062e-05,
+      "loss": 0.8525,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936461204874318e-05,
+      "loss": 0.8477,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931969007146988e-05,
+      "loss": 0.3797,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927474114985186e-05,
+      "loss": 0.8493,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692297653013606e-05,
+      "loss": 0.8268,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69184762543478e-05,
+      "loss": 0.8098,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913973289369657e-05,
+      "loss": 0.3704,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690946763695191e-05,
+      "loss": 0.7736,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69049592988459e-05,
+      "loss": 0.8124,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900448276804004e-05,
+      "loss": 0.7909,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895934572579637e-05,
+      "loss": 0.8084,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6891418187927265e-05,
+      "loss": 0.9127,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886899124602392e-05,
+      "loss": 0.8053,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882377384361562e-05,
+      "loss": 0.7675,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877852968962363e-05,
+      "loss": 0.7993,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687332588016342e-05,
+      "loss": 0.7729,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868796119724396e-05,
+      "loss": 0.8362,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864263689406004e-05,
+      "loss": 0.7921,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685972859096997e-05,
+      "loss": 0.8868,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855190826179088e-05,
+      "loss": 0.8748,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850650396797163e-05,
+      "loss": 0.803,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846107304589047e-05,
+      "loss": 0.8296,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6841561551320628e-05,
+      "loss": 0.8627,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837013138758824e-05,
+      "loss": 0.7828,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832462068671588e-05,
+      "loss": 0.8708,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682790834282791e-05,
+      "loss": 0.8897,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823351962997807e-05,
+      "loss": 0.7599,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6818792930952327e-05,
+      "loss": 0.8961,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814231248463556e-05,
+      "loss": 0.8578,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809666917304603e-05,
+      "loss": 0.3651,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680509993924961e-05,
+      "loss": 0.8428,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800530316073748e-05,
+      "loss": 0.7897,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6795958049553208e-05,
+      "loss": 0.8517,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6791383141465222e-05,
+      "loss": 0.833,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786805593588045e-05,
+      "loss": 0.7853,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678222540770095e-05,
+      "loss": 0.8681,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777642585584243e-05,
+      "loss": 0.8246,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677305712901925e-05,
+      "loss": 0.8122,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676846903978833e-05,
+      "loss": 0.387,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763878319674853e-05,
+      "loss": 0.8138,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759284970463215e-05,
+      "loss": 0.8113,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6754688993938843e-05,
+      "loss": 0.9065,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750090391888175e-05,
+      "loss": 0.8152,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745489166098672e-05,
+      "loss": 0.8467,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740885318358824e-05,
+      "loss": 0.7663,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736278850458122e-05,
+      "loss": 0.8057,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731669764187087e-05,
+      "loss": 0.8555,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727058061337264e-05,
+      "loss": 0.8409,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722443743701194e-05,
+      "loss": 0.7987,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671782681307246e-05,
+      "loss": 0.8429,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671320727124565e-05,
+      "loss": 0.8393,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708585120016353e-05,
+      "loss": 0.799,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670396036118119e-05,
+      "loss": 0.84,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699332996537795e-05,
+      "loss": 0.8761,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694703027884802e-05,
+      "loss": 0.8682,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669007045702187e-05,
+      "loss": 0.8095,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668543528574967e-05,
+      "loss": 0.8852,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680797515869868e-05,
+      "loss": 0.3911,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6676157149185152e-05,
+      "loss": 0.8027,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671514187499226e-05,
+      "loss": 0.8157,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666868632616786e-05,
+      "loss": 0.8088,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6662220486343553e-05,
+      "loss": 0.7893,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657569750486238e-05,
+      "loss": 0.7081,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665291642685257e-05,
+      "loss": 0.8367,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648260517251286e-05,
+      "loss": 0.8528,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643602023492116e-05,
+      "loss": 0.8714,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638940947385806e-05,
+      "loss": 0.841,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634277290744105e-05,
+      "loss": 0.7928,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629611055379754e-05,
+      "loss": 0.862,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662494224310651e-05,
+      "loss": 0.86,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662027085573912e-05,
+      "loss": 0.8406,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615596895093342e-05,
+      "loss": 0.3036,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610920362985932e-05,
+      "loss": 0.8481,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606241261234637e-05,
+      "loss": 0.8194,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660155959165821e-05,
+      "loss": 0.8971,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596875356076403e-05,
+      "loss": 0.3517,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659218855630997e-05,
+      "loss": 0.8591,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587499194180644e-05,
+      "loss": 0.8377,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582807271511172e-05,
+      "loss": 0.8062,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657811279012529e-05,
+      "loss": 0.9211,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573415751847722e-05,
+      "loss": 0.8387,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568716158504194e-05,
+      "loss": 0.8269,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656401401192143e-05,
+      "loss": 0.833,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655930931392713e-05,
+      "loss": 0.8565,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6554602066350003e-05,
+      "loss": 0.7533,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654989227101974e-05,
+      "loss": 0.7772,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545179929767017e-05,
+      "loss": 0.7939,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6540465044423513e-05,
+      "loss": 0.8082,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535747616821884e-05,
+      "loss": 0.8603,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531027648795784e-05,
+      "loss": 0.831,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652630514217985e-05,
+      "loss": 0.3415,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65215800988097e-05,
+      "loss": 0.9119,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516852520521946e-05,
+      "loss": 0.94,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512122409154184e-05,
+      "loss": 0.845,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507389766544997e-05,
+      "loss": 0.8173,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502654594533938e-05,
+      "loss": 0.8111,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649791689496156e-05,
+      "loss": 0.809,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493176669669396e-05,
+      "loss": 0.8542,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648843392049995e-05,
+      "loss": 0.8445,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6483688649296714e-05,
+      "loss": 0.9017,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478940857904163e-05,
+      "loss": 0.8271,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6474190548167744e-05,
+      "loss": 0.7393,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469437721933892e-05,
+      "loss": 0.8549,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6464682381050015e-05,
+      "loss": 0.8212,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459924527364496e-05,
+      "loss": 0.8833,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64551641627267e-05,
+      "loss": 0.8195,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645040128898697e-05,
+      "loss": 0.817,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445635907996612e-05,
+      "loss": 0.8157,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440868021607915e-05,
+      "loss": 0.8989,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436097631674148e-05,
+      "loss": 0.3333,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643132474004954e-05,
+      "loss": 0.8723,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64265493485893e-05,
+      "loss": 0.8018,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421771459149615e-05,
+      "loss": 0.8515,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6416991073587628e-05,
+      "loss": 0.7818,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641220819376146e-05,
+      "loss": 0.8586,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407422821530207e-05,
+      "loss": 0.874,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640263495875392e-05,
+      "loss": 0.8541,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639784460729363e-05,
+      "loss": 0.7242,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6393051769011333e-05,
+      "loss": 0.802,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388256445769987e-05,
+      "loss": 0.835,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383458639433522e-05,
+      "loss": 0.8893,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637865835186683e-05,
+      "loss": 0.8361,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373855584935765e-05,
+      "loss": 0.8365,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369050340507152e-05,
+      "loss": 0.8232,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364242620448767e-05,
+      "loss": 0.8516,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6359432426629363e-05,
+      "loss": 0.7936,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635461976091864e-05,
+      "loss": 0.8581,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349804625187275e-05,
+      "loss": 0.9114,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634498702130689e-05,
+      "loss": 0.8066,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634016695115007e-05,
+      "loss": 0.3866,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633534441659037e-05,
+      "loss": 0.8234,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633051941950228e-05,
+      "loss": 0.8574,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632569196176128e-05,
+      "loss": 0.8243,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632086204524377e-05,
+      "loss": 0.8365,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316029671827137e-05,
+      "loss": 0.7705,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311194843389697e-05,
+      "loss": 0.8168,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306357561810743e-05,
+      "loss": 0.9095,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63015178289705e-05,
+      "loss": 0.8014,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6296675646750168e-05,
+      "loss": 0.8426,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629183101703188e-05,
+      "loss": 0.8,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286983941698737e-05,
+      "loss": 0.8146,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282134422634778e-05,
+      "loss": 0.802,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627728246172499e-05,
+      "loss": 0.8285,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6272428060855316e-05,
+      "loss": 0.87,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626757122191265e-05,
+      "loss": 0.8655,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262711946784835e-05,
+      "loss": 0.7987,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257850237360647e-05,
+      "loss": 0.8073,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625298609552982e-05,
+      "loss": 0.805,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6248119523183032e-05,
+      "loss": 0.8119,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243250522211903e-05,
+      "loss": 0.8296,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6238379094509e-05,
+      "loss": 0.8743,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623350524196783e-05,
+      "loss": 0.8095,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228628966482844e-05,
+      "loss": 0.7819,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6223750269949433e-05,
+      "loss": 0.8385,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218869154263934e-05,
+      "loss": 0.8463,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213985621323624e-05,
+      "loss": 0.8297,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6209099673026706e-05,
+      "loss": 0.9165,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620421131127235e-05,
+      "loss": 0.899,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199320537960627e-05,
+      "loss": 0.8124,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194427354992578e-05,
+      "loss": 0.8658,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189531764270167e-05,
+      "loss": 0.8475,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184633767696292e-05,
+      "loss": 0.8684,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617973336717479e-05,
+      "loss": 0.7669,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174830564610426e-05,
+      "loss": 0.7779,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169925361908912e-05,
+      "loss": 0.8147,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165017760976878e-05,
+      "loss": 0.8243,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61601077637219e-05,
+      "loss": 0.8294,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155195372052473e-05,
+      "loss": 0.3411,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615028058787803e-05,
+      "loss": 0.8425,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145363413108927e-05,
+      "loss": 0.8869,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6140443849656466e-05,
+      "loss": 0.8229,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135521899432854e-05,
+      "loss": 0.8217,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130597564351245e-05,
+      "loss": 0.8165,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125670846325707e-05,
+      "loss": 0.8679,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612074174727125e-05,
+      "loss": 0.8813,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115810269103783e-05,
+      "loss": 0.7786,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110876413740168e-05,
+      "loss": 0.8575,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105940183098176e-05,
+      "loss": 0.8574,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101001579096504e-05,
+      "loss": 0.9648,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096060603654775e-05,
+      "loss": 0.787,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6091117258693524e-05,
+      "loss": 0.8321,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086171546134216e-05,
+      "loss": 0.8415,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081223467899236e-05,
+      "loss": 0.3589,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6076273025911886e-05,
+      "loss": 0.8596,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071320222096387e-05,
+      "loss": 0.8563,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066365058377877e-05,
+      "loss": 0.8258,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606140753668241e-05,
+      "loss": 0.8623,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605644765893697e-05,
+      "loss": 0.8703,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051485427069434e-05,
+      "loss": 0.8216,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604652084300861e-05,
+      "loss": 0.8262,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041553908684214e-05,
+      "loss": 0.8508,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603658462602688e-05,
+      "loss": 0.8674,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031612996968157e-05,
+      "loss": 0.7807,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602663902344049e-05,
+      "loss": 0.3663,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602166270737726e-05,
+      "loss": 0.8368,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016684050712735e-05,
+      "loss": 0.8293,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011703055382107e-05,
+      "loss": 0.7881,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6006719723321477e-05,
+      "loss": 0.7543,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001734056467844e-05,
+      "loss": 0.7997,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996746056759123e-05,
+      "loss": 0.8023,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5991755726134134e-05,
+      "loss": 0.8391,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986763066532608e-05,
+      "loss": 0.3236,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598176807989517e-05,
+      "loss": 0.8401,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976770768163352e-05,
+      "loss": 0.874,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59717711332796e-05,
+      "loss": 0.7691,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5966769177187254e-05,
+      "loss": 0.7907,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961764901830563e-05,
+      "loss": 0.8378,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5956758309154664e-05,
+      "loss": 0.8993,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595174940110561e-05,
+      "loss": 0.835,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594673817963035e-05,
+      "loss": 0.7583,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941724646676725e-05,
+      "loss": 0.8552,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936708804193478e-05,
+      "loss": 0.8764,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931690654130255e-05,
+      "loss": 0.7958,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5926670198437596e-05,
+      "loss": 0.8544,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5921647439066933e-05,
+      "loss": 0.8137,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59166223779706e-05,
+      "loss": 0.9203,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5911595017101817e-05,
+      "loss": 0.7741,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590656535841471e-05,
+      "loss": 0.7726,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901533403864286e-05,
+      "loss": 0.7358,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896499155406453e-05,
+      "loss": 0.8269,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589146261499801e-05,
+      "loss": 0.8686,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5886423784596635e-05,
+      "loss": 0.3431,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881382666160917e-05,
+      "loss": 0.8216,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876339261650313e-05,
+      "loss": 0.8065,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5871293573025188e-05,
+      "loss": 0.7738,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866245602246775e-05,
+      "loss": 0.8933,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586119535127721e-05,
+      "loss": 0.7848,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585614282207951e-05,
+      "loss": 0.9033,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851088016617575e-05,
+      "loss": 0.8368,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5846030936856193e-05,
+      "loss": 0.8481,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5840971584761037e-05,
+      "loss": 0.8017,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5835909962298654e-05,
+      "loss": 0.7967,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830846071436492e-05,
+      "loss": 0.809,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582577991414286e-05,
+      "loss": 0.7625,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5820711492386963e-05,
+      "loss": 0.8636,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5815640808138876e-05,
+      "loss": 0.7735,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581056786336956e-05,
+      "loss": 0.8339,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805492660050852e-05,
+      "loss": 0.3797,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580041520015547e-05,
+      "loss": 0.8746,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795335485657003e-05,
+      "loss": 0.7714,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790253518529923e-05,
+      "loss": 0.8074,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785169300749576e-05,
+      "loss": 0.8421,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578008283429217e-05,
+      "loss": 0.8559,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5774994121134816e-05,
+      "loss": 0.8879,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769903163255468e-05,
+      "loss": 0.8382,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576480996263297e-05,
+      "loss": 0.8179,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759714521247036e-05,
+      "loss": 0.8006,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575461684107824e-05,
+      "loss": 0.8218,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574951692410804e-05,
+      "loss": 0.8663,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744414772318756e-05,
+      "loss": 0.8113,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5739310387693578e-05,
+      "loss": 0.888,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5734203772216564e-05,
+      "loss": 0.8082,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729094927872642e-05,
+      "loss": 0.7595,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5723983856647595e-05,
+      "loss": 0.8671,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5718870560528096e-05,
+      "loss": 0.8727,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5713755041501652e-05,
+      "loss": 0.8901,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708637301556657e-05,
+      "loss": 0.8415,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5703517342682357e-05,
+      "loss": 0.8091,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698395166868867e-05,
+      "loss": 0.7993,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693270776107157e-05,
+      "loss": 0.7946,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688144172389067e-05,
+      "loss": 0.8186,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5683015357707282e-05,
+      "loss": 0.8006,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567788433405537e-05,
+      "loss": 0.814,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672751103427733e-05,
+      "loss": 0.8357,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566761566781964e-05,
+      "loss": 0.855,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662478029227225e-05,
+      "loss": 0.3471,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657338189647472e-05,
+      "loss": 0.8447,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652196151078218e-05,
+      "loss": 0.8373,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647051915518154e-05,
+      "loss": 0.8342,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641905484966834e-05,
+      "loss": 0.8765,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636756861424656e-05,
+      "loss": 0.7797,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5631606046892877e-05,
+      "loss": 0.7996,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5626453043373598e-05,
+      "loss": 0.8219,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5621297852869778e-05,
+      "loss": 0.7988,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561614047738522e-05,
+      "loss": 0.806,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610980918924586e-05,
+      "loss": 0.8185,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605819179493374e-05,
+      "loss": 0.8265,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600655261097938e-05,
+      "loss": 0.8559,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5595489165745476e-05,
+      "loss": 0.8125,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5590320895444035e-05,
+      "loss": 0.8249,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5585150452202502e-05,
+      "loss": 0.8224,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579977838030617e-05,
+      "loss": 0.8003,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557480305493895e-05,
+      "loss": 0.8288,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569626104938934e-05,
+      "loss": 0.8551,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556444699004283e-05,
+      "loss": 0.833,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559265712263737e-05,
+      "loss": 0.8966,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5554082273615608e-05,
+      "loss": 0.8875,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5548896676113223e-05,
+      "loss": 0.822,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5543708921772217e-05,
+      "loss": 0.7922,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538519012609043e-05,
+      "loss": 0.8325,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553332695064101e-05,
+      "loss": 0.82,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528132737886252e-05,
+      "loss": 0.8545,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5522936376363747e-05,
+      "loss": 0.78,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55177378680933e-05,
+      "loss": 0.8385,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551253721509556e-05,
+      "loss": 0.8282,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507334419392e-05,
+      "loss": 0.8305,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502129483004927e-05,
+      "loss": 0.8086,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5496922407957492e-05,
+      "loss": 0.7824,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549171319627366e-05,
+      "loss": 0.8003,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486501849978244e-05,
+      "loss": 0.8095,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548128837109687e-05,
+      "loss": 0.8596,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547607276165601e-05,
+      "loss": 0.7883,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547085502368294e-05,
+      "loss": 0.8506,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465635159205798e-05,
+      "loss": 0.7451,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546041317025351e-05,
+      "loss": 0.84,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5455189058855858e-05,
+      "loss": 0.8595,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5449962827043433e-05,
+      "loss": 0.848,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5444734476847658e-05,
+      "loss": 0.8221,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439504010300776e-05,
+      "loss": 0.8422,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543427142943585e-05,
+      "loss": 0.7945,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5429036736286767e-05,
+      "loss": 0.8019,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542379993288824e-05,
+      "loss": 0.7901,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418561021275792e-05,
+      "loss": 0.8373,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541332000348578e-05,
+      "loss": 0.3394,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408076881555364e-05,
+      "loss": 0.7875,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540283165752253e-05,
+      "loss": 0.7759,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397584333426083e-05,
+      "loss": 0.3412,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539233491130564e-05,
+      "loss": 0.8741,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5387083393201633e-05,
+      "loss": 0.8285,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5381829781155307e-05,
+      "loss": 0.7971,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5376574077208737e-05,
+      "loss": 0.785,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371316283404786e-05,
+      "loss": 0.8787,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536605640178715e-05,
+      "loss": 0.7714,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360794434400322e-05,
+      "loss": 0.3496,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355530383289616e-05,
+      "loss": 0.8826,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350264250501154e-05,
+      "loss": 0.834,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534499603808186e-05,
+      "loss": 0.3484,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533972574807947e-05,
+      "loss": 0.8093,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533445338254254e-05,
+      "loss": 0.8327,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5329178943520407e-05,
+      "loss": 0.7833,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5323902433063246e-05,
+      "loss": 0.8098,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5318623853222006e-05,
+      "loss": 0.7605,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313343206048458e-05,
+      "loss": 0.7699,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308060493595173e-05,
+      "loss": 0.8345,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5302775717915528e-05,
+      "loss": 0.8225,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5297488881063696e-05,
+      "loss": 0.7813,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292199985094652e-05,
+      "loss": 0.8532,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286909032064174e-05,
+      "loss": 0.8585,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5281616024028844e-05,
+      "loss": 0.8882,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527632096304603e-05,
+      "loss": 0.3913,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271023851173908e-05,
+      "loss": 0.8519,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5265724690471454e-05,
+      "loss": 0.8307,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260423482998427e-05,
+      "loss": 0.8421,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5255120230815394e-05,
+      "loss": 0.8318,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5249814935983713e-05,
+      "loss": 0.3219,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244507600565532e-05,
+      "loss": 0.8387,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239198226623796e-05,
+      "loss": 0.8715,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5233886816222247e-05,
+      "loss": 0.7927,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522857337142541e-05,
+      "loss": 0.8016,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223257894298602e-05,
+      "loss": 0.8175,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217940386907933e-05,
+      "loss": 0.8566,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212620851320305e-05,
+      "loss": 0.849,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.52072992896034e-05,
+      "loss": 0.8132,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201975703825698e-05,
+      "loss": 0.8266,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196650096056451e-05,
+      "loss": 0.8697,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191322468365716e-05,
+      "loss": 0.8384,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185992822824318e-05,
+      "loss": 0.7776,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180661161503877e-05,
+      "loss": 0.8547,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175327486476792e-05,
+      "loss": 0.836,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169991799816247e-05,
+      "loss": 0.7777,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164654103596204e-05,
+      "loss": 0.7585,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159314399891406e-05,
+      "loss": 0.8053,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153972690777386e-05,
+      "loss": 0.8577,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148628978330444e-05,
+      "loss": 0.7454,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514328326462767e-05,
+      "loss": 0.7869,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513793555174692e-05,
+      "loss": 0.368,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132585841766835e-05,
+      "loss": 0.8833,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127234136766832e-05,
+      "loss": 0.8071,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121880438827101e-05,
+      "loss": 0.8182,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116524750028607e-05,
+      "loss": 0.8137,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111167072453091e-05,
+      "loss": 0.857,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105807408183059e-05,
+      "loss": 0.8358,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100445759301805e-05,
+      "loss": 0.796,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095082127893377e-05,
+      "loss": 0.786,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508971651604261e-05,
+      "loss": 0.3249,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508434892583509e-05,
+      "loss": 0.7535,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078979359357194e-05,
+      "loss": 0.7503,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073607818696043e-05,
+      "loss": 0.7802,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068234305939546e-05,
+      "loss": 0.8645,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062858823176374e-05,
+      "loss": 0.8614,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057481372495948e-05,
+      "loss": 0.8413,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052101955988478e-05,
+      "loss": 0.7916,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5046720575744917e-05,
+      "loss": 0.3417,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041337233856998e-05,
+      "loss": 0.696,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035951932417205e-05,
+      "loss": 0.3557,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503056467351879e-05,
+      "loss": 0.8014,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5025175459255761e-05,
+      "loss": 0.8152,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019784291722888e-05,
+      "loss": 0.8177,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50143911730157e-05,
+      "loss": 0.762,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008996105230493e-05,
+      "loss": 0.8575,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003599090464307e-05,
+      "loss": 0.7966,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998200130814941e-05,
+      "loss": 0.8333,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992799228380959e-05,
+      "loss": 0.8702,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987396385261676e-05,
+      "loss": 0.8008,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981991603557155e-05,
+      "loss": 0.8228,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976584885368221e-05,
+      "loss": 0.7834,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4971176232796447e-05,
+      "loss": 0.8116,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4965765647944161e-05,
+      "loss": 0.8017,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960353132914436e-05,
+      "loss": 0.7744,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954938689811105e-05,
+      "loss": 0.8286,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494952232073874e-05,
+      "loss": 0.8233,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944104027802668e-05,
+      "loss": 0.8583,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938683813108968e-05,
+      "loss": 0.8124,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493326167876445e-05,
+      "loss": 0.8186,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492783762687669e-05,
+      "loss": 0.7706,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922411659553993e-05,
+      "loss": 0.8744,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916983778905418e-05,
+      "loss": 0.8816,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911553987040767e-05,
+      "loss": 0.7532,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906122286070582e-05,
+      "loss": 0.7509,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900688678106146e-05,
+      "loss": 0.8694,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4895253165259487e-05,
+      "loss": 0.8889,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4889815749643371e-05,
+      "loss": 0.7779,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884376433371304e-05,
+      "loss": 0.8744,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878935218557531e-05,
+      "loss": 0.3732,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873492107317039e-05,
+      "loss": 0.8073,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868047101765541e-05,
+      "loss": 0.8097,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862600204019497e-05,
+      "loss": 0.8117,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857151416196103e-05,
+      "loss": 0.8348,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485170074041328e-05,
+      "loss": 0.8047,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484624817878969e-05,
+      "loss": 0.8398,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840793733444727e-05,
+      "loss": 0.7816,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835337406498516e-05,
+      "loss": 0.7957,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4829879200071916e-05,
+      "loss": 0.7953,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824419116286514e-05,
+      "loss": 0.9071,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818957157264624e-05,
+      "loss": 0.875,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813493325129296e-05,
+      "loss": 0.8609,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48080276220043e-05,
+      "loss": 0.7676,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4802560050014145e-05,
+      "loss": 0.7771,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797090611284053e-05,
+      "loss": 0.818,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479161930793998e-05,
+      "loss": 0.7877,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786146142108601e-05,
+      "loss": 0.8411,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780671115917324e-05,
+      "loss": 0.8373,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477519423149427e-05,
+      "loss": 0.8145,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769715490968289e-05,
+      "loss": 0.7323,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764234896468946e-05,
+      "loss": 0.7774,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758752450126536e-05,
+      "loss": 0.7742,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753268154072067e-05,
+      "loss": 0.8699,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747782010437264e-05,
+      "loss": 0.8466,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4742294021354578e-05,
+      "loss": 0.7348,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736804188957174e-05,
+      "loss": 0.7899,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473131251537893e-05,
+      "loss": 0.82,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472581900275444e-05,
+      "loss": 0.7779,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4720323653219017e-05,
+      "loss": 0.7875,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471482646890869e-05,
+      "loss": 0.7689,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709327451960191e-05,
+      "loss": 0.8602,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703826604510978e-05,
+      "loss": 0.3429,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698323928699203e-05,
+      "loss": 0.772,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692819426663744e-05,
+      "loss": 0.853,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687313100544188e-05,
+      "loss": 0.7932,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681804952480817e-05,
+      "loss": 0.8231,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4676294984614641e-05,
+      "loss": 0.8253,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467078319908736e-05,
+      "loss": 0.8522,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466526959804139e-05,
+      "loss": 0.8169,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659754183619847e-05,
+      "loss": 0.8207,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654236957966558e-05,
+      "loss": 0.8635,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648717923226057e-05,
+      "loss": 0.8609,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643197081543566e-05,
+      "loss": 0.8071,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637674435065024e-05,
+      "loss": 0.8413,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632149985937063e-05,
+      "loss": 0.7749,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462662373630702e-05,
+      "loss": 0.774,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621095688322935e-05,
+      "loss": 0.3587,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615565844133539e-05,
+      "loss": 0.7769,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610034205888263e-05,
+      "loss": 0.8137,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604500775737241e-05,
+      "loss": 0.7701,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598965555831297e-05,
+      "loss": 0.8374,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593428548321957e-05,
+      "loss": 0.9031,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587889755361437e-05,
+      "loss": 0.8251,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4582349179102643e-05,
+      "loss": 0.7714,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576806821699188e-05,
+      "loss": 0.7509,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571262685305365e-05,
+      "loss": 0.8598,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565716772076161e-05,
+      "loss": 0.8114,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560169084167259e-05,
+      "loss": 0.7633,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554619623735023e-05,
+      "loss": 0.816,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549068392936514e-05,
+      "loss": 0.9089,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543515393929472e-05,
+      "loss": 0.3576,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537960628872338e-05,
+      "loss": 0.8273,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453240409992423e-05,
+      "loss": 0.8445,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526845809244952e-05,
+      "loss": 0.8079,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4521285758994994e-05,
+      "loss": 0.8055,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515723951335527e-05,
+      "loss": 0.3505,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510160388428414e-05,
+      "loss": 0.8102,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4504595072436192e-05,
+      "loss": 0.7728,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449902800552208e-05,
+      "loss": 0.8621,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493459189849976e-05,
+      "loss": 0.8308,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448788862758447e-05,
+      "loss": 0.8467,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482316320890815e-05,
+      "loss": 0.8553,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476742271934954e-05,
+      "loss": 0.8102,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44711664828835e-05,
+      "loss": 0.8178,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465588955903745e-05,
+      "loss": 0.8754,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460009693163651e-05,
+      "loss": 0.871,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454428696831868e-05,
+      "loss": 0.7934,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448845969077706e-05,
+      "loss": 0.8969,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443261512071155e-05,
+      "loss": 0.8227,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443767532798288e-05,
+      "loss": 0.792,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432087418984207e-05,
+      "loss": 0.7863,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426497787247143e-05,
+      "loss": 0.8279,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420906434944355e-05,
+      "loss": 0.8092,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415313364249188e-05,
+      "loss": 0.7957,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409718577335654e-05,
+      "loss": 0.8105,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404122076378425e-05,
+      "loss": 0.8246,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439852386355284e-05,
+      "loss": 0.8281,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439292394103491e-05,
+      "loss": 0.3683,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387322311001308e-05,
+      "loss": 0.8505,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381718975629373e-05,
+      "loss": 0.8805,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376113937097098e-05,
+      "loss": 0.8325,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370507197583145e-05,
+      "loss": 0.824,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364898759266837e-05,
+      "loss": 0.8459,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359288624328158e-05,
+      "loss": 0.8695,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353676794947747e-05,
+      "loss": 0.7523,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348063273306905e-05,
+      "loss": 0.7705,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342448061587587e-05,
+      "loss": 0.9226,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4336831161972418e-05,
+      "loss": 0.9111,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331212576644657e-05,
+      "loss": 0.7833,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432559230778824e-05,
+      "loss": 0.7974,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319970357587742e-05,
+      "loss": 0.702,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43143467282284e-05,
+      "loss": 0.748,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43087214218961e-05,
+      "loss": 0.8349,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430309444077738e-05,
+      "loss": 0.8026,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297465787059427e-05,
+      "loss": 0.8075,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429183546293009e-05,
+      "loss": 0.8413,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286203470577853e-05,
+      "loss": 0.8295,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280569812191852e-05,
+      "loss": 0.8272,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427493448996187e-05,
+      "loss": 0.7693,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269297506078348e-05,
+      "loss": 0.8438,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426365886273236e-05,
+      "loss": 0.8611,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258018562115623e-05,
+      "loss": 0.8609,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425237660642051e-05,
+      "loss": 0.7804,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424673299784003e-05,
+      "loss": 0.8629,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4241087738567837e-05,
+      "loss": 0.7937,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423544083079822e-05,
+      "loss": 0.8448,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229792276726122e-05,
+      "loss": 0.8365,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224142078547112e-05,
+      "loss": 0.7671,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4218490238457407e-05,
+      "loss": 0.8267,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212836758653862e-05,
+      "loss": 0.8799,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207181641333962e-05,
+      "loss": 0.3798,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201524888695835e-05,
+      "loss": 0.7924,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195866502938245e-05,
+      "loss": 0.8349,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190206486260585e-05,
+      "loss": 0.8146,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184544840862888e-05,
+      "loss": 0.8922,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178881568945816e-05,
+      "loss": 0.8003,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173216672710665e-05,
+      "loss": 0.7599,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167550154359365e-05,
+      "loss": 0.8596,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161882016094475e-05,
+      "loss": 0.8442,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156212260119176e-05,
+      "loss": 0.8196,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150540888637287e-05,
+      "loss": 0.7512,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144867903853254e-05,
+      "loss": 0.8181,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139193307972148e-05,
+      "loss": 0.8702,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133517103199666e-05,
+      "loss": 0.8002,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127839291742129e-05,
+      "loss": 0.8391,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412215987580649e-05,
+      "loss": 0.8685,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116478857600311e-05,
+      "loss": 0.8209,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110796239331791e-05,
+      "loss": 0.83,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410511202320975e-05,
+      "loss": 0.8848,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4099426211443619e-05,
+      "loss": 0.7577,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093738806243458e-05,
+      "loss": 0.8129,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088049809819939e-05,
+      "loss": 0.8575,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4082359224384361e-05,
+      "loss": 0.3471,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076667052148635e-05,
+      "loss": 0.7971,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070973295325294e-05,
+      "loss": 0.7626,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065277956127478e-05,
+      "loss": 0.7944,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405958103676895e-05,
+      "loss": 0.8115,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053882539464083e-05,
+      "loss": 0.8177,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404818246642787e-05,
+      "loss": 0.8238,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042480819875907e-05,
+      "loss": 0.3536,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036777602024405e-05,
+      "loss": 0.7955,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031072815090188e-05,
+      "loss": 0.8399,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402536646129069e-05,
+      "loss": 0.7965,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019658542843952e-05,
+      "loss": 0.75,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013949061968626e-05,
+      "loss": 0.8325,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008238020883965e-05,
+      "loss": 0.8545,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4002525421809831e-05,
+      "loss": 0.8606,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996811266966704e-05,
+      "loss": 0.8357,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991095558575645e-05,
+      "loss": 0.3397,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985378298858341e-05,
+      "loss": 0.801,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979659490037065e-05,
+      "loss": 0.8102,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397393913433471e-05,
+      "loss": 0.7738,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396821723397475e-05,
+      "loss": 0.8351,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962493791181275e-05,
+      "loss": 0.813,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395676880817897e-05,
+      "loss": 0.7937,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951042287193116e-05,
+      "loss": 0.8788,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945314230449594e-05,
+      "loss": 0.8304,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393958464017488e-05,
+      "loss": 0.7999,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933853518596048e-05,
+      "loss": 0.8038,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392812086794077e-05,
+      "loss": 0.838,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922386690437307e-05,
+      "loss": 0.7854,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3916650988314516e-05,
+      "loss": 0.8343,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910913763801843e-05,
+      "loss": 0.8126,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905175019129338e-05,
+      "loss": 0.8434,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899434756527623e-05,
+      "loss": 0.8515,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893692978227924e-05,
+      "loss": 0.7836,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887949686462055e-05,
+      "loss": 0.8096,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388220488346241e-05,
+      "loss": 0.796,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387645857146198e-05,
+      "loss": 0.8379,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870710752694334e-05,
+      "loss": 0.7977,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864961429393632e-05,
+      "loss": 0.8177,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859210603794626e-05,
+      "loss": 0.3271,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853458278132637e-05,
+      "loss": 0.7826,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3847704454643571e-05,
+      "loss": 0.8432,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841949135563928e-05,
+      "loss": 0.7934,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836192323130782e-05,
+      "loss": 0.8186,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830434019581786e-05,
+      "loss": 0.7752,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824674227155175e-05,
+      "loss": 0.8149,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381891294808976e-05,
+      "loss": 0.8169,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813150184624936e-05,
+      "loss": 0.7566,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807385939000668e-05,
+      "loss": 0.3663,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38016202134575e-05,
+      "loss": 0.8175,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795853010236555e-05,
+      "loss": 0.7507,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379008433157952e-05,
+      "loss": 0.7687,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784314179728668e-05,
+      "loss": 0.8669,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377854255692684e-05,
+      "loss": 0.8162,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377276946541744e-05,
+      "loss": 0.8251,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376699490744446e-05,
+      "loss": 0.7643,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761218885252443e-05,
+      "loss": 0.8804,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755441401086517e-05,
+      "loss": 0.7886,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374966245719237e-05,
+      "loss": 0.3607,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743882055816257e-05,
+      "loss": 0.716,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738100199205007e-05,
+      "loss": 0.8522,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732316889606005e-05,
+      "loss": 0.7847,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726532129267206e-05,
+      "loss": 0.8797,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720745920437128e-05,
+      "loss": 0.8431,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714958265364855e-05,
+      "loss": 0.3433,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709169166300028e-05,
+      "loss": 0.8395,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703378625492852e-05,
+      "loss": 0.8723,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369758664519409e-05,
+      "loss": 0.8598,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691793227655066e-05,
+      "loss": 0.7987,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685998375127669e-05,
+      "loss": 0.7985,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680202089864331e-05,
+      "loss": 0.8203,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367440437411806e-05,
+      "loss": 0.8834,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668605230142403e-05,
+      "loss": 0.8736,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662804660191469e-05,
+      "loss": 0.7846,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657002666519925e-05,
+      "loss": 0.7858,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651199251382983e-05,
+      "loss": 0.8773,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645394417036414e-05,
+      "loss": 0.7934,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639588165736544e-05,
+      "loss": 0.8118,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633780499740237e-05,
+      "loss": 0.8034,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627971421304916e-05,
+      "loss": 0.8001,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622160932688553e-05,
+      "loss": 0.8074,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616349036149668e-05,
+      "loss": 0.8151,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610535733947325e-05,
+      "loss": 0.7944,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604721028341137e-05,
+      "loss": 0.8486,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598904921591261e-05,
+      "loss": 0.7948,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593087415958397e-05,
+      "loss": 0.8127,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587268513703795e-05,
+      "loss": 0.8416,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581448217089243e-05,
+      "loss": 0.7678,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575626528377073e-05,
+      "loss": 0.8725,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569803449830152e-05,
+      "loss": 0.9022,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563978983711899e-05,
+      "loss": 0.7904,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355815313228626e-05,
+      "loss": 0.8733,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552325897817728e-05,
+      "loss": 0.8085,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546497282571331e-05,
+      "loss": 0.3707,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3540667288812633e-05,
+      "loss": 0.8553,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534835918807731e-05,
+      "loss": 0.8122,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529003174823265e-05,
+      "loss": 0.7847,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35231690591264e-05,
+      "loss": 0.8121,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517333573984842e-05,
+      "loss": 0.7758,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511496721666823e-05,
+      "loss": 0.758,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350565850444111e-05,
+      "loss": 0.8353,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499818924577e-05,
+      "loss": 0.7542,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493977984344317e-05,
+      "loss": 0.7698,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.348813568601342e-05,
+      "loss": 0.8281,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348229203185519e-05,
+      "loss": 0.7771,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476447024141038e-05,
+      "loss": 0.761,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470600665142894e-05,
+      "loss": 0.7891,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464752957133226e-05,
+      "loss": 0.8291,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458903902385018e-05,
+      "loss": 0.8014,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453053503171782e-05,
+      "loss": 0.7551,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447201761767543e-05,
+      "loss": 0.8,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344134868044686e-05,
+      "loss": 0.7678,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435494261484804e-05,
+      "loss": 0.8088,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342963850715697e-05,
+      "loss": 0.3499,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3423781419739472e-05,
+      "loss": 0.8033,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341792300150894e-05,
+      "loss": 0.8507,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3412063254742527e-05,
+      "loss": 0.8086,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340620218171789e-05,
+      "loss": 0.3576,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340033978471322e-05,
+      "loss": 0.8073,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394476066007202e-05,
+      "loss": 0.8054,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388611027879052e-05,
+      "loss": 0.8108,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382744672608489e-05,
+      "loss": 0.7665,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376877002475746e-05,
+      "loss": 0.8312,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3371008019761568e-05,
+      "loss": 0.7412,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365137726747213e-05,
+      "loss": 0.792,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359266125714443e-05,
+      "loss": 0.7825,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353393218945532e-05,
+      "loss": 0.8141,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347519008723258e-05,
+      "loss": 0.8071,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341643497330911e-05,
+      "loss": 0.7616,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333576668705228e-05,
+      "loss": 0.8183,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329888580171666e-05,
+      "loss": 0.8042,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324009178973872e-05,
+      "loss": 0.3469,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318128485744197e-05,
+      "loss": 0.7561,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312246502768453e-05,
+      "loss": 0.7464,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306363232332944e-05,
+      "loss": 0.7732,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300478676724482e-05,
+      "loss": 0.742,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294592838230375e-05,
+      "loss": 0.8685,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288705719138429e-05,
+      "loss": 0.8684,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3282817321736946e-05,
+      "loss": 0.8203,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276927648314732e-05,
+      "loss": 0.8856,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271036701161082e-05,
+      "loss": 0.798,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326514448256579e-05,
+      "loss": 0.8245,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259250994819141e-05,
+      "loss": 0.8234,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253356240211917e-05,
+      "loss": 0.8513,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324746022103539e-05,
+      "loss": 0.8632,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324156293958132e-05,
+      "loss": 0.8094,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235664398141968e-05,
+      "loss": 0.8239,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229764599010077e-05,
+      "loss": 0.8219,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223863544478875e-05,
+      "loss": 0.8228,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217961236842089e-05,
+      "loss": 0.8433,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212057678393926e-05,
+      "loss": 0.7952,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320615287142908e-05,
+      "loss": 0.8034,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200246818242733e-05,
+      "loss": 0.8294,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194339521130544e-05,
+      "loss": 0.865,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188430982388664e-05,
+      "loss": 0.7855,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3182521204313726e-05,
+      "loss": 0.8513,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176610189202836e-05,
+      "loss": 0.7703,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170697939353593e-05,
+      "loss": 0.8016,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164784457064067e-05,
+      "loss": 0.8268,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158869744632807e-05,
+      "loss": 0.7392,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315295380435885e-05,
+      "loss": 0.786,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147036638541698e-05,
+      "loss": 0.8332,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141118249481335e-05,
+      "loss": 0.7822,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135198639478222e-05,
+      "loss": 0.8161,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129277810833292e-05,
+      "loss": 0.7794,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123355765847952e-05,
+      "loss": 0.7889,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117432506824083e-05,
+      "loss": 0.8099,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111508036064037e-05,
+      "loss": 0.8316,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105582355870636e-05,
+      "loss": 0.8768,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099655468547174e-05,
+      "loss": 0.7897,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093727376397413e-05,
+      "loss": 0.3836,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087798081725587e-05,
+      "loss": 0.8307,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081867586836388e-05,
+      "loss": 0.779,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075935894034984e-05,
+      "loss": 0.783,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070003005627008e-05,
+      "loss": 0.8478,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064068923918551e-05,
+      "loss": 0.789,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058133651216172e-05,
+      "loss": 0.7546,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052197189826892e-05,
+      "loss": 0.7488,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046259542058198e-05,
+      "loss": 0.8258,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040320710218035e-05,
+      "loss": 0.8144,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034380696614806e-05,
+      "loss": 0.7914,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028439503557376e-05,
+      "loss": 0.7571,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022497133355066e-05,
+      "loss": 0.3564,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016553588317662e-05,
+      "loss": 0.7991,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3010608870755395e-05,
+      "loss": 0.8118,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004662982978964e-05,
+      "loss": 0.8109,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998715927299513e-05,
+      "loss": 0.7995,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992767706028641e-05,
+      "loss": 0.8156,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298681832147841e-05,
+      "loss": 0.8217,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980867775961324e-05,
+      "loss": 0.8775,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974916071790337e-05,
+      "loss": 0.8282,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2968963211278865e-05,
+      "loss": 0.818,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2963009196740762e-05,
+      "loss": 0.7596,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957054030490332e-05,
+      "loss": 0.7401,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951097714842331e-05,
+      "loss": 0.8297,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945140252111967e-05,
+      "loss": 0.8157,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293918164461488e-05,
+      "loss": 0.7843,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293322189466716e-05,
+      "loss": 0.8489,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927261004585347e-05,
+      "loss": 0.8261,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921298976686421e-05,
+      "loss": 0.8113,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915335813287802e-05,
+      "loss": 0.8078,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909371516707355e-05,
+      "loss": 0.782,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903406089263375e-05,
+      "loss": 0.8496,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897439533274615e-05,
+      "loss": 0.8248,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891471851060251e-05,
+      "loss": 0.7226,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885503044939903e-05,
+      "loss": 0.7967,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879533117233628e-05,
+      "loss": 0.8401,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2873562070261918e-05,
+      "loss": 0.8052,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28675899063457e-05,
+      "loss": 0.745,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2861616627806334e-05,
+      "loss": 0.8631,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855642236965615e-05,
+      "loss": 0.7745,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849666736145774e-05,
+      "loss": 0.8304,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284369012766946e-05,
+      "loss": 0.8547,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837712413859773e-05,
+      "loss": 0.822,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831733597040224e-05,
+      "loss": 0.8312,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825753679534762e-05,
+      "loss": 0.7362,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819772663667762e-05,
+      "loss": 0.8457,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281379055176403e-05,
+      "loss": 0.3496,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807807346148786e-05,
+      "loss": 0.8242,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801823049147689e-05,
+      "loss": 0.8368,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795837663086812e-05,
+      "loss": 0.8216,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789851190292664e-05,
+      "loss": 0.8361,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783863633092159e-05,
+      "loss": 0.8671,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777874993812646e-05,
+      "loss": 0.7715,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277188527478189e-05,
+      "loss": 0.8101,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765894478328073e-05,
+      "loss": 0.7933,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759902606779804e-05,
+      "loss": 0.7797,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753909662466099e-05,
+      "loss": 0.7719,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2747915647716403e-05,
+      "loss": 0.8198,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741920564860562e-05,
+      "loss": 0.7558,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735924416228853e-05,
+      "loss": 0.79,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2729927204151958e-05,
+      "loss": 0.8241,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723928930960972e-05,
+      "loss": 0.7889,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717929598987406e-05,
+      "loss": 0.7947,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711929210563187e-05,
+      "loss": 0.765,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705927768020638e-05,
+      "loss": 0.8229,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699925273692507e-05,
+      "loss": 0.8204,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269392172991194e-05,
+      "loss": 0.8456,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26879171390125e-05,
+      "loss": 0.8372,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681911503328153e-05,
+      "loss": 0.8039,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675904825193264e-05,
+      "loss": 0.8639,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669897106942615e-05,
+      "loss": 0.8147,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2663888350911389e-05,
+      "loss": 0.8324,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657878559435166e-05,
+      "loss": 0.7298,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2651867734849935e-05,
+      "loss": 0.8474,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645855879492084e-05,
+      "loss": 0.8325,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639842995698402e-05,
+      "loss": 0.788,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263382908580608e-05,
+      "loss": 0.7816,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627814152152702e-05,
+      "loss": 0.8462,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621798197076259e-05,
+      "loss": 0.8106,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261578122291513e-05,
+      "loss": 0.7704,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609763232008092e-05,
+      "loss": 0.751,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2603744226694322e-05,
+      "loss": 0.8035,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597724209313392e-05,
+      "loss": 0.7758,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591703182205257e-05,
+      "loss": 0.7858,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585681147710276e-05,
+      "loss": 0.8547,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257965810816919e-05,
+      "loss": 0.8455,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257363406592314e-05,
+      "loss": 0.7609,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567609023313648e-05,
+      "loss": 0.813,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2561582982682634e-05,
+      "loss": 0.7032,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25555559463724e-05,
+      "loss": 0.8114,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549527916725632e-05,
+      "loss": 0.7709,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2543498896085409e-05,
+      "loss": 0.8435,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253746888679519e-05,
+      "loss": 0.8487,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531437891198825e-05,
+      "loss": 0.8455,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2525405911640542e-05,
+      "loss": 0.8531,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251937295046495e-05,
+      "loss": 0.7935,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251333901001704e-05,
+      "loss": 0.7973,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507304092642191e-05,
+      "loss": 0.8007,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501268200686153e-05,
+      "loss": 0.7791,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249523133649506e-05,
+      "loss": 0.3463,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248919350241542e-05,
+      "loss": 0.8205,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483154700794122e-05,
+      "loss": 0.8308,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477114933978429e-05,
+      "loss": 0.7702,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471074204315975e-05,
+      "loss": 0.8709,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465032514154778e-05,
+      "loss": 0.7958,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458989865843223e-05,
+      "loss": 0.8174,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452946261730069e-05,
+      "loss": 0.7936,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446901704164441e-05,
+      "loss": 0.7894,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244085619549584e-05,
+      "loss": 0.8096,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434809738074144e-05,
+      "loss": 0.866,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428762334249584e-05,
+      "loss": 0.8414,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422713986372772e-05,
+      "loss": 0.8713,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416664696794675e-05,
+      "loss": 0.8114,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241061446786664e-05,
+      "loss": 0.8383,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404563301940366e-05,
+      "loss": 0.8533,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239851120136793e-05,
+      "loss": 0.8043,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392458168501755e-05,
+      "loss": 0.7113,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238640420569464e-05,
+      "loss": 0.7834,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380349315299744e-05,
+      "loss": 0.8119,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237429349967058e-05,
+      "loss": 0.7464,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368236761161023e-05,
+      "loss": 0.7725,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362179102125315e-05,
+      "loss": 0.8088,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356120524918045e-05,
+      "loss": 0.7953,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2350061031894159e-05,
+      "loss": 0.7466,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344000625408964e-05,
+      "loss": 0.7769,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337939307818127e-05,
+      "loss": 0.7612,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331877081477661e-05,
+      "loss": 0.7599,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232581394874393e-05,
+      "loss": 0.7827,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2319749911973658e-05,
+      "loss": 0.7747,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313684973523914e-05,
+      "loss": 0.82,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307619135752123e-05,
+      "loss": 0.8097,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2301552401016058e-05,
+      "loss": 0.797,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295484771673838e-05,
+      "loss": 0.8131,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228941625008393e-05,
+      "loss": 0.7988,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283346838605148e-05,
+      "loss": 0.8258,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227727653959666e-05,
+      "loss": 0.7523,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2271205355417966e-05,
+      "loss": 0.7481,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265133288428919e-05,
+      "loss": 0.8374,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225906034098971e-05,
+      "loss": 0.7217,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252986515460878e-05,
+      "loss": 0.3348,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2246911814203295e-05,
+      "loss": 0.7414,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224083623957818e-05,
+      "loss": 0.7792,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2234759793947092e-05,
+      "loss": 0.8992,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228682479671927e-05,
+      "loss": 0.8485,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222604299114911e-05,
+      "loss": 0.782,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216525254638618e-05,
+      "loss": 0.8713,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210445348605954e-05,
+      "loss": 0.845,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204364583380164e-05,
+      "loss": 0.8025,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219828296132481e-05,
+      "loss": 0.7649,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192200484803812e-05,
+      "loss": 0.7052,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186117156181401e-05,
+      "loss": 0.7669,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2180032977822149e-05,
+      "loss": 0.7583,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173947952090961e-05,
+      "loss": 0.7974,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167862081353063e-05,
+      "loss": 0.8536,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2161775367974019e-05,
+      "loss": 0.7751,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155687814319705e-05,
+      "loss": 0.7538,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149599422756347e-05,
+      "loss": 0.783,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143510195650473e-05,
+      "loss": 0.8385,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137420135368951e-05,
+      "loss": 0.7857,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131329244278969e-05,
+      "loss": 0.7763,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125237524748037e-05,
+      "loss": 0.8012,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119144979143984e-05,
+      "loss": 0.8326,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113051609834966e-05,
+      "loss": 0.7945,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106957419189453e-05,
+      "loss": 0.7819,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100862409576243e-05,
+      "loss": 0.768,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094766583364445e-05,
+      "loss": 0.7647,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088669942923486e-05,
+      "loss": 0.7692,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082572490623113e-05,
+      "loss": 0.8428,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2076474228833383e-05,
+      "loss": 0.7791,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070375159924674e-05,
+      "loss": 0.8702,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064275286267675e-05,
+      "loss": 0.8372,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058174610233384e-05,
+      "loss": 0.7946,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052073134193118e-05,
+      "loss": 0.8234,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045970860518504e-05,
+      "loss": 0.8459,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203986779158147e-05,
+      "loss": 0.8676,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033763929754265e-05,
+      "loss": 0.8589,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2027659277409436e-05,
+      "loss": 0.3669,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2021553836919848e-05,
+      "loss": 0.8225,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015447610658661e-05,
+      "loss": 0.8399,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200934060099935e-05,
+      "loss": 0.7786,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2003232810315688e-05,
+      "loss": 0.7781,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997124240981754e-05,
+      "loss": 0.3514,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199101489537193e-05,
+      "loss": 0.749,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984904775860897e-05,
+      "loss": 0.8273,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978793884823643e-05,
+      "loss": 0.7925,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972682224635454e-05,
+      "loss": 0.7807,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196656979767191e-05,
+      "loss": 0.8804,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960456606308887e-05,
+      "loss": 0.8319,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195434265292257e-05,
+      "loss": 0.8229,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194822793988943e-05,
+      "loss": 0.7974,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942112469586241e-05,
+      "loss": 0.7334,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935996244390064e-05,
+      "loss": 0.3766,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1929879266678256e-05,
+      "loss": 0.8506,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923761538828465e-05,
+      "loss": 0.3396,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917643063218638e-05,
+      "loss": 0.7433,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911523842227002e-05,
+      "loss": 0.8656,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190540387823208e-05,
+      "loss": 0.8566,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899283173612686e-05,
+      "loss": 0.8735,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893161730747917e-05,
+      "loss": 0.8052,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887039552017157e-05,
+      "loss": 0.8215,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1880916639800076e-05,
+      "loss": 0.7463,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1874792996476634e-05,
+      "loss": 0.7969,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186866862442707e-05,
+      "loss": 0.7474,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862543526031914e-05,
+      "loss": 0.7633,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856417703671962e-05,
+      "loss": 0.8064,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1850291159728307e-05,
+      "loss": 0.82,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1844163896582321e-05,
+      "loss": 0.8129,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838035916615645e-05,
+      "loss": 0.8233,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831907222210212e-05,
+      "loss": 0.7128,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182577781574822e-05,
+      "loss": 0.8518,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819647699612153e-05,
+      "loss": 0.9295,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181351687618477e-05,
+      "loss": 0.7929,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1807385347849098e-05,
+      "loss": 0.3793,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1801253116988447e-05,
+      "loss": 0.7901,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179512018598639e-05,
+      "loss": 0.8244,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788986557226788e-05,
+      "loss": 0.8375,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782852233093754e-05,
+      "loss": 0.7267,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177671721597168e-05,
+      "loss": 0.7777,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770581508245235e-05,
+      "loss": 0.7446,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764445112299347e-05,
+      "loss": 0.7206,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758308030519208e-05,
+      "loss": 0.8161,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752170265290287e-05,
+      "loss": 0.782,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746031818998314e-05,
+      "loss": 0.3712,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739892694029285e-05,
+      "loss": 0.8136,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733752892769457e-05,
+      "loss": 0.8263,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727612417605348e-05,
+      "loss": 0.7806,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1721471270923746e-05,
+      "loss": 0.8198,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171532945511169e-05,
+      "loss": 0.7767,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709186972556494e-05,
+      "loss": 0.8078,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703043825645715e-05,
+      "loss": 0.7329,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696900016767173e-05,
+      "loss": 0.7858,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169075554830895e-05,
+      "loss": 0.7516,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684610422659382e-05,
+      "loss": 0.7842,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678464642207057e-05,
+      "loss": 0.3168,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1672318209340824e-05,
+      "loss": 0.7649,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666171126449779e-05,
+      "loss": 0.8635,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660023395923279e-05,
+      "loss": 0.7575,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165387502015092e-05,
+      "loss": 0.8457,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647726001522562e-05,
+      "loss": 0.8183,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641576342428305e-05,
+      "loss": 0.7813,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1635426045258508e-05,
+      "loss": 0.868,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1629275112403766e-05,
+      "loss": 0.7648,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623123546254931e-05,
+      "loss": 0.823,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616971349203091e-05,
+      "loss": 0.7822,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1610818523639598e-05,
+      "loss": 0.7628,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1604665071956023e-05,
+      "loss": 0.8336,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598510996544201e-05,
+      "loss": 0.7724,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592356299796198e-05,
+      "loss": 0.8044,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586200984104323e-05,
+      "loss": 0.8038,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580045051861138e-05,
+      "loss": 0.8238,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573888505459423e-05,
+      "loss": 0.7716,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567731347292217e-05,
+      "loss": 0.7329,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156157357975278e-05,
+      "loss": 0.7484,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555415205234623e-05,
+      "loss": 0.7875,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549256226131482e-05,
+      "loss": 0.7552,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543096644837335e-05,
+      "loss": 0.7356,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536936463746395e-05,
+      "loss": 0.8446,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530775685253101e-05,
+      "loss": 0.7519,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524614311752128e-05,
+      "loss": 0.8153,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518452345638383e-05,
+      "loss": 0.8177,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512289789307003e-05,
+      "loss": 0.8505,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506126645153355e-05,
+      "loss": 0.6963,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499962915573035e-05,
+      "loss": 0.7919,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493798602961862e-05,
+      "loss": 0.8132,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1487633709715884e-05,
+      "loss": 0.8285,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481468238231378e-05,
+      "loss": 0.7706,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1475302190904847e-05,
+      "loss": 0.7706,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146913557013301e-05,
+      "loss": 0.8385,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462968378312808e-05,
+      "loss": 0.7417,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456800617841417e-05,
+      "loss": 0.8164,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1450632291116226e-05,
+      "loss": 0.8265,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1444463400534837e-05,
+      "loss": 0.7669,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438293948495086e-05,
+      "loss": 0.8594,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432123937395015e-05,
+      "loss": 0.7188,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142595336963289e-05,
+      "loss": 0.8575,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141978224760719e-05,
+      "loss": 0.7886,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413610573716607e-05,
+      "loss": 0.7874,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140743835036006e-05,
+      "loss": 0.7897,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1401265579936669e-05,
+      "loss": 0.7908,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1395092264845764e-05,
+      "loss": 0.7971,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13889184074869e-05,
+      "loss": 0.7618,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382744010259835e-05,
+      "loss": 0.8842,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376569075564537e-05,
+      "loss": 0.7952,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1370393605801184e-05,
+      "loss": 0.3762,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136421760337016e-05,
+      "loss": 0.8125,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135804107067206e-05,
+      "loss": 0.7696,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351864010107678e-05,
+      "loss": 0.8139,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345686424078027e-05,
+      "loss": 0.8038,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1339508314984306e-05,
+      "loss": 0.808,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333329685227934e-05,
+      "loss": 0.8216,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1327150537210517e-05,
+      "loss": 0.7627,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320970873333876e-05,
+      "loss": 0.8353,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314790696000023e-05,
+      "loss": 0.7375,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308610007611178e-05,
+      "loss": 0.7934,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302428810569755e-05,
+      "loss": 0.8679,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296247107278362e-05,
+      "loss": 0.8797,
+      "step": 2467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290064900139808e-05,
+      "loss": 0.7293,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1283882191557099e-05,
+      "loss": 0.8558,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277698983933431e-05,
+      "loss": 0.8153,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271515279672205e-05,
+      "loss": 0.8011,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265331081177002e-05,
+      "loss": 0.8058,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1259146390851598e-05,
+      "loss": 0.8365,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252961211099966e-05,
+      "loss": 0.7427,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246775544326268e-05,
+      "loss": 0.7975,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1240589392934849e-05,
+      "loss": 0.7476,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123440275933025e-05,
+      "loss": 0.7969,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122821564591719e-05,
+      "loss": 0.773,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222028055100584e-05,
+      "loss": 0.7172,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215839989285536e-05,
+      "loss": 0.8316,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209651450877315e-05,
+      "loss": 0.8031,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203462442281394e-05,
+      "loss": 0.8742,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119727296590342e-05,
+      "loss": 0.8072,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1191083024149225e-05,
+      "loss": 0.8776,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184892619424815e-05,
+      "loss": 0.8407,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117870175413638e-05,
+      "loss": 0.6456,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1172510430690298e-05,
+      "loss": 0.8057,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116631865149311e-05,
+      "loss": 0.8449,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160126418951542e-05,
+      "loss": 0.7798,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153933735472498e-05,
+      "loss": 0.7361,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147740603463051e-05,
+      "loss": 0.7724,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1141547025330454e-05,
+      "loss": 0.8238,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1135353003482132e-05,
+      "loss": 0.8337,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112915854032568e-05,
+      "loss": 0.8333,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122963638268864e-05,
+      "loss": 0.8557,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116768299719628e-05,
+      "loss": 0.7967,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110572527086079e-05,
+      "loss": 0.7809,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104376322776493e-05,
+      "loss": 0.8474,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098179689199316e-05,
+      "loss": 0.8345,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091982628763155e-05,
+      "loss": 0.7933,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10857851438768e-05,
+      "loss": 0.3716,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107958723694918e-05,
+      "loss": 0.8197,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073388910389411e-05,
+      "loss": 0.8092,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1067190166606758e-05,
+      "loss": 0.8043,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060991008010658e-05,
+      "loss": 0.8868,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10547914370107e-05,
+      "loss": 0.834,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048591456016638e-05,
+      "loss": 0.7984,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042391067438385e-05,
+      "loss": 0.3591,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1036190273686019e-05,
+      "loss": 0.8363,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029989077169761e-05,
+      "loss": 0.8478,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023787480299996e-05,
+      "loss": 0.8234,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101758548548727e-05,
+      "loss": 0.7883,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011383095142279e-05,
+      "loss": 0.7846,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005180311675869e-05,
+      "loss": 0.8397,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0998977137499044e-05,
+      "loss": 0.7494,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0992773575022962e-05,
+      "loss": 0.7776,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0986569626658918e-05,
+      "loss": 0.8188,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0980365294818377e-05,
+      "loss": 0.8079,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097416058191294e-05,
+      "loss": 0.3374,
+      "step": 2519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096795549035436e-05,
+      "loss": 0.8315,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961750022554533e-05,
+      "loss": 0.7764,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095554418092551e-05,
+      "loss": 0.7674,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949337967879478e-05,
+      "loss": 0.786,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943131385828773e-05,
+      "loss": 0.7887,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0936924437185879e-05,
+      "loss": 0.7778,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093071712436341e-05,
+      "loss": 0.8471,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092450944977413e-05,
+      "loss": 0.7826,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918301415830945e-05,
+      "loss": 0.8552,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09120930249469e-05,
+      "loss": 0.8356,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905884279535174e-05,
+      "loss": 0.7715,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0899675182009086e-05,
+      "loss": 0.78,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893465734782094e-05,
+      "loss": 0.7719,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887255940267792e-05,
+      "loss": 0.7911,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881045800879904e-05,
+      "loss": 0.8057,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0874835319032295e-05,
+      "loss": 0.8034,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086862449713896e-05,
+      "loss": 0.8003,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862413337614021e-05,
+      "loss": 0.8208,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0856201842871741e-05,
+      "loss": 0.7687,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0849990015326508e-05,
+      "loss": 0.3418,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843777857392836e-05,
+      "loss": 0.8474,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837565371485375e-05,
+      "loss": 0.7702,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831352560018897e-05,
+      "loss": 0.804,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825139425408304e-05,
+      "loss": 0.7953,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818925970068617e-05,
+      "loss": 0.7566,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812712196414989e-05,
+      "loss": 0.8039,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08064981068627e-05,
+      "loss": 0.8214,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080028370382714e-05,
+      "loss": 0.7917,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079406898972383e-05,
+      "loss": 0.8397,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787853966968408e-05,
+      "loss": 0.792,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0781638637976636e-05,
+      "loss": 0.8228,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775423005164394e-05,
+      "loss": 0.7604,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0769207070947675e-05,
+      "loss": 0.8054,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762990837742598e-05,
+      "loss": 0.8111,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756774307965385e-05,
+      "loss": 0.3411,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075055748403239e-05,
+      "loss": 0.8192,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744340368360069e-05,
+      "loss": 0.349,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0738122963364996e-05,
+      "loss": 0.7795,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731905271463858e-05,
+      "loss": 0.3509,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725687295073449e-05,
+      "loss": 0.7969,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071946903661068e-05,
+      "loss": 0.7944,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0713250498492565e-05,
+      "loss": 0.8203,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707031683136232e-05,
+      "loss": 0.3411,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700812592958918e-05,
+      "loss": 0.8158,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0694593230377959e-05,
+      "loss": 0.7779,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688373597810801e-05,
+      "loss": 0.7985,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682153697675e-05,
+      "loss": 0.815,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067593353238821e-05,
+      "loss": 0.7772,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669713104368189e-05,
+      "loss": 0.7921,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663492416032799e-05,
+      "loss": 0.7851,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657271469799998e-05,
+      "loss": 0.813,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651050268087853e-05,
+      "loss": 0.795,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064482881331452e-05,
+      "loss": 0.7491,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638607107898266e-05,
+      "loss": 0.7899,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632385154257445e-05,
+      "loss": 0.7873,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626162954810508e-05,
+      "loss": 0.8089,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619940511976006e-05,
+      "loss": 0.7373,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613717828172584e-05,
+      "loss": 0.7407,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060749490581898e-05,
+      "loss": 0.7647,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601271747334025e-05,
+      "loss": 0.7734,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595048355136634e-05,
+      "loss": 0.8056,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058882473164583e-05,
+      "loss": 0.7669,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582600879280709e-05,
+      "loss": 0.8631,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0576376800460464e-05,
+      "loss": 0.7778,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570152497604378e-05,
+      "loss": 0.7648,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0563927973131816e-05,
+      "loss": 0.738,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557703229462229e-05,
+      "loss": 0.8578,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0551478269015154e-05,
+      "loss": 0.7378,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054525309421022e-05,
+      "loss": 0.8245,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0539027707467126e-05,
+      "loss": 0.8422,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0532802111205667e-05,
+      "loss": 0.7804,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0526576307845706e-05,
+      "loss": 0.723,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520350299807196e-05,
+      "loss": 0.8122,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514124089510167e-05,
+      "loss": 0.7952,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507897679374729e-05,
+      "loss": 0.8264,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0501671071821064e-05,
+      "loss": 0.7972,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495444269269434e-05,
+      "loss": 0.8335,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0489217274140185e-05,
+      "loss": 0.8119,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482990088853721e-05,
+      "loss": 0.8394,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476762715830538e-05,
+      "loss": 0.7351,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0470535157491189e-05,
+      "loss": 0.7822,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046430741625631e-05,
+      "loss": 0.7713,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458079494546605e-05,
+      "loss": 0.8029,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0451851394782844e-05,
+      "loss": 0.767,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044562311938587e-05,
+      "loss": 0.7713,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0439394670776598e-05,
+      "loss": 0.7445,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0433166051376007e-05,
+      "loss": 0.7594,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426937263605135e-05,
+      "loss": 0.7717,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420708309885096e-05,
+      "loss": 0.7944,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414479192637065e-05,
+      "loss": 0.8421,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040824991428228e-05,
+      "loss": 0.802,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402020477242043e-05,
+      "loss": 0.8521,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395790883937718e-05,
+      "loss": 0.3498,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038956113679072e-05,
+      "loss": 0.7579,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383331238222539e-05,
+      "loss": 0.7857,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0377101190654716e-05,
+      "loss": 0.786,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370870996508849e-05,
+      "loss": 0.8259,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03646406582066e-05,
+      "loss": 0.7562,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358410178169673e-05,
+      "loss": 0.739,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352179558819844e-05,
+      "loss": 0.8386,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345948802578932e-05,
+      "loss": 0.7595,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339717911868808e-05,
+      "loss": 0.8133,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0333486889111408e-05,
+      "loss": 0.8531,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0327255736728707e-05,
+      "loss": 0.8118,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321024457142732e-05,
+      "loss": 0.8271,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0314793052775563e-05,
+      "loss": 0.7931,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308561526049327e-05,
+      "loss": 0.7988,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302329879386203e-05,
+      "loss": 0.7853,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296098115208408e-05,
+      "loss": 0.8722,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289866235938208e-05,
+      "loss": 0.7648,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283634243997917e-05,
+      "loss": 0.8034,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0277402141809891e-05,
+      "loss": 0.8108,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027116993179653e-05,
+      "loss": 0.3654,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264937616380274e-05,
+      "loss": 0.7468,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0258705197983598e-05,
+      "loss": 0.797,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252472679029028e-05,
+      "loss": 0.3534,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0246240061939128e-05,
+      "loss": 0.8121,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0240007349136489e-05,
+      "loss": 0.7599,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233774543043753e-05,
+      "loss": 0.755,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022754164608359e-05,
+      "loss": 0.776,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0221308660678708e-05,
+      "loss": 0.7755,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215075589251846e-05,
+      "loss": 0.7991,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020884243422578e-05,
+      "loss": 0.8442,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202609198023324e-05,
+      "loss": 0.7696,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196375883067309e-05,
+      "loss": 0.8029,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0190142491780608e-05,
+      "loss": 0.8438,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0183909026586118e-05,
+      "loss": 0.7399,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177675489906773e-05,
+      "loss": 0.349,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171441884165524e-05,
+      "loss": 0.7613,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016520821178536e-05,
+      "loss": 0.7518,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015897447518928e-05,
+      "loss": 0.7624,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152740676800326e-05,
+      "loss": 0.8279,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014650681904155e-05,
+      "loss": 0.7823,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014027290433604e-05,
+      "loss": 0.8561,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134038935106892e-05,
+      "loss": 0.3404,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127804913777235e-05,
+      "loss": 0.8589,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012157084277021e-05,
+      "loss": 0.7788,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115336724508984e-05,
+      "loss": 0.7878,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109102561416732e-05,
+      "loss": 0.8284,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0102868355916662e-05,
+      "loss": 0.7973,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096634110431989e-05,
+      "loss": 0.8164,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090399827385942e-05,
+      "loss": 0.76,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084165509201764e-05,
+      "loss": 0.7456,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077931158302718e-05,
+      "loss": 0.8228,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071696777112076e-05,
+      "loss": 0.7586,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065462368053127e-05,
+      "loss": 0.3242,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059227933549163e-05,
+      "loss": 0.7426,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052993476023485e-05,
+      "loss": 0.784,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046758997899409e-05,
+      "loss": 0.8093,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040524501600259e-05,
+      "loss": 0.8002,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034289989549364e-05,
+      "loss": 0.8182,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002805546417006e-05,
+      "loss": 0.784,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0021820927885684e-05,
+      "loss": 0.8431,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001558638311958e-05,
+      "loss": 0.8476,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009351832295103e-05,
+      "loss": 0.7248,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0003117277835593e-05,
+      "loss": 0.7879,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99688272216441e-06,
+      "loss": 0.776,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990648167704904e-06,
+      "loss": 0.7909,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984413616880422e-06,
+      "loss": 0.8017,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97817907211432e-06,
+      "loss": 0.3624,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971944535829943e-06,
+      "loss": 0.733,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965710010450637e-06,
+      "loss": 0.8536,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959475498399743e-06,
+      "loss": 0.8385,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953241002100593e-06,
+      "loss": 0.832,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94700652397652e-06,
+      "loss": 0.7848,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940772066450842e-06,
+      "loss": 0.812,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934537631946875e-06,
+      "loss": 0.7244,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928303222887925e-06,
+      "loss": 0.7909,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922068841697284e-06,
+      "loss": 0.8258,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915834490798239e-06,
+      "loss": 0.7598,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909600172614062e-06,
+      "loss": 0.7512,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903365889568015e-06,
+      "loss": 0.7615,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897131644083338e-06,
+      "loss": 0.7662,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890897438583271e-06,
+      "loss": 0.8585,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884663275491023e-06,
+      "loss": 0.7241,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878429157229792e-06,
+      "loss": 0.7454,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872195086222768e-06,
+      "loss": 0.7061,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86596106489311e-06,
+      "loss": 0.7657,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859727095663962e-06,
+      "loss": 0.7386,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853493180958449e-06,
+      "loss": 0.7937,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847259323199674e-06,
+      "loss": 0.763,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841025524810724e-06,
+      "loss": 0.7922,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834791788214645e-06,
+      "loss": 0.8214,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828558115834477e-06,
+      "loss": 0.3103,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82232451009323e-06,
+      "loss": 0.7414,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816090973413885e-06,
+      "loss": 0.7558,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809857508219395e-06,
+      "loss": 0.7351,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803624116932695e-06,
+      "loss": 0.7758,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79739080197668e-06,
+      "loss": 0.8454,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79115756577422e-06,
+      "loss": 0.8361,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784924410748159e-06,
+      "loss": 0.757,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778691339321297e-06,
+      "loss": 0.776,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.772458353916413e-06,
+      "loss": 0.7873,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766225456956248e-06,
+      "loss": 0.8077,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759992650863513e-06,
+      "loss": 0.8033,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753759938060875e-06,
+      "loss": 0.3152,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747527320970974e-06,
+      "loss": 0.7495,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741294802016403e-06,
+      "loss": 0.7595,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735062383619731e-06,
+      "loss": 0.3448,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728830068203472e-06,
+      "loss": 0.7541,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722597858190112e-06,
+      "loss": 0.8087,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716365756002086e-06,
+      "loss": 0.7757,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710133764061795e-06,
+      "loss": 0.7665,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703901884791594e-06,
+      "loss": 0.7968,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.6976701206138e-06,
+      "loss": 0.7774,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691438473950674e-06,
+      "loss": 0.3276,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.685206947224437e-06,
+      "loss": 0.7628,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678975542857275e-06,
+      "loss": 0.8436,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6727442632713e-06,
+      "loss": 0.7901,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666513110888595e-06,
+      "loss": 0.7563,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660282088131193e-06,
+      "loss": 0.7463,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654051197421073e-06,
+      "loss": 0.7197,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647820441180159e-06,
+      "loss": 0.8593,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641589821830326e-06,
+      "loss": 0.8345,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.635359341793404e-06,
+      "loss": 0.7963,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629129003491153e-06,
+      "loss": 0.7364,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622898809345289e-06,
+      "loss": 0.7655,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616668761777465e-06,
+      "loss": 0.8135,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610438863209284e-06,
+      "loss": 0.873,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604209116062286e-06,
+      "loss": 0.7473,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59797952275796e-06,
+      "loss": 0.7235,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59175008571772e-06,
+      "loss": 0.7872,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585520807362936e-06,
+      "loss": 0.8329,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579291690114903e-06,
+      "loss": 0.7795,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573062736394868e-06,
+      "loss": 0.8061,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566833948623999e-06,
+      "loss": 0.8095,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560605329223406e-06,
+      "loss": 0.77,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554376880614133e-06,
+      "loss": 0.7797,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54814860521716e-06,
+      "loss": 0.8086,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5419205054534e-06,
+      "loss": 0.7673,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535692583743691e-06,
+      "loss": 0.778,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529464842508813e-06,
+      "loss": 0.7651,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523237284169469e-06,
+      "loss": 0.7612,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517009911146282e-06,
+      "loss": 0.7928,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51078272585982e-06,
+      "loss": 0.8508,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504555730730567e-06,
+      "loss": 0.8503,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49832892817894e-06,
+      "loss": 0.8202,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492102320625275e-06,
+      "loss": 0.804,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485875910489835e-06,
+      "loss": 0.8822,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479649700192805e-06,
+      "loss": 0.3632,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473423692154299e-06,
+      "loss": 0.7383,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.467197888794338e-06,
+      "loss": 0.7926,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460972292532876e-06,
+      "loss": 0.8272,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454746905789785e-06,
+      "loss": 0.7681,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448521730984848e-06,
+      "loss": 0.7443,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442296770537775e-06,
+      "loss": 0.7407,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436072026868186e-06,
+      "loss": 0.8211,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.429847502395623e-06,
+      "loss": 0.7609,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.423623199539536e-06,
+      "loss": 0.7681,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417399120719296e-06,
+      "loss": 0.7706,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.411175268354174e-06,
+      "loss": 0.7969,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404951644863367e-06,
+      "loss": 0.8212,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398728252665978e-06,
+      "loss": 0.7907,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392505094181023e-06,
+      "loss": 0.8594,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386282171827417e-06,
+      "loss": 0.7417,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.380059488023995e-06,
+      "loss": 0.8206,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373837045189494e-06,
+      "loss": 0.779,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36761484574256e-06,
+      "loss": 0.8684,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361392892101735e-06,
+      "loss": 0.6905,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.355171186685483e-06,
+      "loss": 0.7367,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34894973191215e-06,
+      "loss": 0.81,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342728530200005e-06,
+      "loss": 0.8354,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336507583967204e-06,
+      "loss": 0.7561,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330286895631814e-06,
+      "loss": 0.7797,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.324066467611792e-06,
+      "loss": 0.7745,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317846302325e-06,
+      "loss": 0.7794,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311626402189202e-06,
+      "loss": 0.8144,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305406769622046e-06,
+      "loss": 0.843,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299187407041087e-06,
+      "loss": 0.7547,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29296831686377e-06,
+      "loss": 0.8218,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286749501507438e-06,
+      "loss": 0.766,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280530963389324e-06,
+      "loss": 0.7754,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274312704926553e-06,
+      "loss": 0.8225,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268094728536144e-06,
+      "loss": 0.8157,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261877036635007e-06,
+      "loss": 0.8584,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255659631639934e-06,
+      "loss": 0.7774,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249442515967614e-06,
+      "loss": 0.8537,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243225692034617e-06,
+      "loss": 0.8243,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237009162257407e-06,
+      "loss": 0.8068,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230792929052326e-06,
+      "loss": 0.7403,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22457699483561e-06,
+      "loss": 0.7939,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218361362023365e-06,
+      "loss": 0.7774,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212146033031593e-06,
+      "loss": 0.8487,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205931010276174e-06,
+      "loss": 0.3438,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199716296172864e-06,
+      "loss": 0.8457,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193501893137302e-06,
+      "loss": 0.8223,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187287803585013e-06,
+      "loss": 0.7673,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181074029931386e-06,
+      "loss": 0.7602,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1748605745917e-06,
+      "loss": 0.8125,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168647439981104e-06,
+      "loss": 0.8186,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162434628514627e-06,
+      "loss": 0.8228,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156222142607167e-06,
+      "loss": 0.3655,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150009984673497e-06,
+      "loss": 0.8056,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143798157128262e-06,
+      "loss": 0.7595,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13758666238598e-06,
+      "loss": 0.8623,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131375502861042e-06,
+      "loss": 0.7936,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.125164680967706e-06,
+      "loss": 0.3457,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118954199120096e-06,
+      "loss": 0.7994,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11274405973221e-06,
+      "loss": 0.7443,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10653426521791e-06,
+      "loss": 0.8108,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100324817990917e-06,
+      "loss": 0.3517,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09411572046483e-06,
+      "loss": 0.7339,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087906975053105e-06,
+      "loss": 0.3309,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081698584169058e-06,
+      "loss": 0.3241,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075490550225873e-06,
+      "loss": 0.8093,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069282875636593e-06,
+      "loss": 0.7728,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063075562814125e-06,
+      "loss": 0.801,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056868614171227e-06,
+      "loss": 0.8169,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050662032120527e-06,
+      "loss": 0.8194,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044455819074493e-06,
+      "loss": 0.8117,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.038249977445469e-06,
+      "loss": 0.8289,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032044509645643e-06,
+      "loss": 0.7803,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025839418087063e-06,
+      "loss": 0.7694,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019634705181624e-06,
+      "loss": 0.8173,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013430373341082e-06,
+      "loss": 0.7809,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00722642497704e-06,
+      "loss": 0.8296,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001022862500959e-06,
+      "loss": 0.7797,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994819688324134e-06,
+      "loss": 0.3511,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988616904857726e-06,
+      "loss": 0.7992,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982414514512732e-06,
+      "loss": 0.8263,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976212519700007e-06,
+      "loss": 0.7098,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970010922830244e-06,
+      "loss": 0.7538,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963809726313983e-06,
+      "loss": 0.8024,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957608932561615e-06,
+      "loss": 0.7282,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951408543983363e-06,
+      "loss": 0.7828,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945208562989306e-06,
+      "loss": 0.7138,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939008991989347e-06,
+      "loss": 0.8448,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.932809833393245e-06,
+      "loss": 0.7734,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926611089610592e-06,
+      "loss": 0.7163,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920412763050823e-06,
+      "loss": 0.3571,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.914214856123205e-06,
+      "loss": 0.7759,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908017371236843e-06,
+      "loss": 0.7239,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901820310800686e-06,
+      "loss": 0.8,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.895623677223512e-06,
+      "loss": 0.746,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889427472913925e-06,
+      "loss": 0.7624,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883231700280376e-06,
+      "loss": 0.7882,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877036361731138e-06,
+      "loss": 0.7863,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870841459674325e-06,
+      "loss": 0.7422,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864646996517871e-06,
+      "loss": 0.3678,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858452974669547e-06,
+      "loss": 0.329,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85225939653695e-06,
+      "loss": 0.7848,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846066264527504e-06,
+      "loss": 0.8064,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839873581048462e-06,
+      "loss": 0.8129,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.833681348506894e-06,
+      "loss": 0.7818,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.827489569309705e-06,
+      "loss": 0.7766,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821298245863621e-06,
+      "loss": 0.8185,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81510738057519e-06,
+      "loss": 0.8126,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808916975850778e-06,
+      "loss": 0.7427,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80272703409658e-06,
+      "loss": 0.6869,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796537557718608e-06,
+      "loss": 0.725,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790348549122689e-06,
+      "loss": 0.7493,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78416001071447e-06,
+      "loss": 0.7518,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777971944899417e-06,
+      "loss": 0.7454,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771784354082812e-06,
+      "loss": 0.7626,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765597240669755e-06,
+      "loss": 0.7601,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759410607065154e-06,
+      "loss": 0.77,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753224455673734e-06,
+      "loss": 0.8117,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747038788900033e-06,
+      "loss": 0.7844,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740853609148402e-06,
+      "loss": 0.8499,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734668918823003e-06,
+      "loss": 0.7936,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728484720327797e-06,
+      "loss": 0.7839,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72230101606657e-06,
+      "loss": 0.7924,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.716117808442905e-06,
+      "loss": 0.8267,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709935099860194e-06,
+      "loss": 0.7771,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70375289272164e-06,
+      "loss": 0.8124,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697571189430247e-06,
+      "loss": 0.7648,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.691389992388822e-06,
+      "loss": 0.8004,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68520930399998e-06,
+      "loss": 0.7638,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679029126666129e-06,
+      "loss": 0.8007,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.672849462789487e-06,
+      "loss": 0.7698,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66667031477207e-06,
+      "loss": 0.752,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660491685015697e-06,
+      "loss": 0.8069,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654313575921975e-06,
+      "loss": 0.3585,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648135989892322e-06,
+      "loss": 0.7979,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641958929327942e-06,
+      "loss": 0.7638,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635782396629844e-06,
+      "loss": 0.7828,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62960639419882e-06,
+      "loss": 0.7499,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623430924435465e-06,
+      "loss": 0.8515,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.617255989740169e-06,
+      "loss": 0.7908,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611081592513103e-06,
+      "loss": 0.7475,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604907735154239e-06,
+      "loss": 0.8358,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598734420063335e-06,
+      "loss": 0.7653,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592561649639942e-06,
+      "loss": 0.7684,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586389426283393e-06,
+      "loss": 0.8134,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580217752392817e-06,
+      "loss": 0.772,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574046630367114e-06,
+      "loss": 0.8237,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567876062604989e-06,
+      "loss": 0.7883,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561706051504917e-06,
+      "loss": 0.8415,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.555536599465166e-06,
+      "loss": 0.8049,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.549367708883777e-06,
+      "loss": 0.8074,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543199382158583e-06,
+      "loss": 0.7696,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53703162168719e-06,
+      "loss": 0.3403,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530864429866996e-06,
+      "loss": 0.829,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.524697809095156e-06,
+      "loss": 0.7144,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518531761768623e-06,
+      "loss": 0.7407,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512366290284119e-06,
+      "loss": 0.8035,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506201397038142e-06,
+      "loss": 0.8019,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500037084426968e-06,
+      "loss": 0.7499,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493873354846647e-06,
+      "loss": 0.793,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487710210692999e-06,
+      "loss": 0.7686,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481547654361619e-06,
+      "loss": 0.7605,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475385688247877e-06,
+      "loss": 0.7482,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.469224314746904e-06,
+      "loss": 0.7863,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463063536253607e-06,
+      "loss": 0.7338,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456903355162667e-06,
+      "loss": 0.8536,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450743773868521e-06,
+      "loss": 0.8549,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44458479476538e-06,
+      "loss": 0.8231,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438426420247221e-06,
+      "loss": 0.7414,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432268652707787e-06,
+      "loss": 0.7581,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426111494540578e-06,
+      "loss": 0.811,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419954948138867e-06,
+      "loss": 0.7551,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413799015895679e-06,
+      "loss": 0.834,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.407643700203806e-06,
+      "loss": 0.8102,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401489003455802e-06,
+      "loss": 0.8158,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395334928043982e-06,
+      "loss": 0.8249,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389181476360407e-06,
+      "loss": 0.3444,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383028650796908e-06,
+      "loss": 0.7784,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37687645374507e-06,
+      "loss": 0.7844,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370724887596237e-06,
+      "loss": 0.7725,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364573954741495e-06,
+      "loss": 0.779,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358423657571698e-06,
+      "loss": 0.7924,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352273998477442e-06,
+      "loss": 0.785,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.346124979849082e-06,
+      "loss": 0.3453,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339976604076724e-06,
+      "loss": 0.7654,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33382887355022e-06,
+      "loss": 0.3066,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327681790659179e-06,
+      "loss": 0.8382,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321535357792947e-06,
+      "loss": 0.8108,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315389577340625e-06,
+      "loss": 0.7757,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309244451691055e-06,
+      "loss": 0.7676,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30309998323283e-06,
+      "loss": 0.773,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296956174354288e-06,
+      "loss": 0.8531,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290813027443508e-06,
+      "loss": 0.7259,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28467054488831e-06,
+      "loss": 0.8081,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.278528729076256e-06,
+      "loss": 0.7702,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272387582394652e-06,
+      "loss": 0.7589,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266247107230548e-06,
+      "loss": 0.7768,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260107305970716e-06,
+      "loss": 0.8181,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253968181001688e-06,
+      "loss": 0.7803,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247829734709716e-06,
+      "loss": 0.8064,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241691969480794e-06,
+      "loss": 0.7414,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235554887700656e-06,
+      "loss": 0.7769,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229418491754766e-06,
+      "loss": 0.7713,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22328278402832e-06,
+      "loss": 0.8096,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.217147766906253e-06,
+      "loss": 0.3579,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211013442773218e-06,
+      "loss": 0.7814,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204879814013612e-06,
+      "loss": 0.7896,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198746883011555e-06,
+      "loss": 0.7811,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.192614652150905e-06,
+      "loss": 0.6993,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186483123815233e-06,
+      "loss": 0.7955,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180352300387849e-06,
+      "loss": 0.7853,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174222184251781e-06,
+      "loss": 0.838,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168092777789793e-06,
+      "loss": 0.8011,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161964083384357e-06,
+      "loss": 0.7078,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155836103417684e-06,
+      "loss": 0.3194,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149708840271694e-06,
+      "loss": 0.3122,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143582296328041e-06,
+      "loss": 0.3276,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13745647396809e-06,
+      "loss": 0.7647,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131331375572931e-06,
+      "loss": 0.7638,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125207003523368e-06,
+      "loss": 0.8488,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119083360199926e-06,
+      "loss": 0.749,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11296044798285e-06,
+      "loss": 0.7424,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106838269252088e-06,
+      "loss": 0.7922,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100716826387316e-06,
+      "loss": 0.7197,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094596121767923e-06,
+      "loss": 0.7199,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088476157773002e-06,
+      "loss": 0.8112,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082356936781365e-06,
+      "loss": 0.7774,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.076238461171536e-06,
+      "loss": 0.817,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.070120733321746e-06,
+      "loss": 0.704,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064003755609941e-06,
+      "loss": 0.799,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057887530413764e-06,
+      "loss": 0.7886,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051772060110572e-06,
+      "loss": 0.7605,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045657347077434e-06,
+      "loss": 0.7471,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039543393691116e-06,
+      "loss": 0.786,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033430202328094e-06,
+      "loss": 0.7442,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02731777536455e-06,
+      "loss": 0.8292,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021206115176357e-06,
+      "loss": 0.7937,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015095224139104e-06,
+      "loss": 0.7769,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008985104628076e-06,
+      "loss": 0.7873,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002875759018251e-06,
+      "loss": 0.7798,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996767189684315e-06,
+      "loss": 0.8037,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990659399000654e-06,
+      "loss": 0.7733,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98455238934134e-06,
+      "loss": 0.6881,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978446163080155e-06,
+      "loss": 0.7736,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972340722590564e-06,
+      "loss": 0.7977,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966236070245737e-06,
+      "loss": 0.7767,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960132208418533e-06,
+      "loss": 0.8212,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954029139481501e-06,
+      "loss": 0.7087,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947926865806884e-06,
+      "loss": 0.7945,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941825389766618e-06,
+      "loss": 0.8464,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935724713732328e-06,
+      "loss": 0.7686,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929624840075329e-06,
+      "loss": 0.8009,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92352577116662e-06,
+      "loss": 0.7488,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917427509376889e-06,
+      "loss": 0.8045,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911330057076514e-06,
+      "loss": 0.3099,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905233416635558e-06,
+      "loss": 0.7344,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899137590423758e-06,
+      "loss": 0.841,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89304258081055e-06,
+      "loss": 0.7216,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886948390165038e-06,
+      "loss": 0.7547,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88085502085602e-06,
+      "loss": 0.7864,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874762475251966e-06,
+      "loss": 0.8693,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868670755721033e-06,
+      "loss": 0.7682,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86257986463105e-06,
+      "loss": 0.8033,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856489804349532e-06,
+      "loss": 0.8088,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.850400577243658e-06,
+      "loss": 0.8425,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844312185680298e-06,
+      "loss": 0.7465,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.838224632025986e-06,
+      "loss": 0.8058,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83213791864694e-06,
+      "loss": 0.8159,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826052047909042e-06,
+      "loss": 0.7819,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819967022177853e-06,
+      "loss": 0.7634,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8138828438186e-06,
+      "loss": 0.7782,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807799515196188e-06,
+      "loss": 0.8055,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801717038675192e-06,
+      "loss": 0.7607,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795635416619843e-06,
+      "loss": 0.8189,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789554651394047e-06,
+      "loss": 0.7846,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783474745361384e-06,
+      "loss": 0.759,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777395700885092e-06,
+      "loss": 0.8249,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771317520328077e-06,
+      "loss": 0.8134,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76524020605291e-06,
+      "loss": 0.7882,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75916376042182e-06,
+      "loss": 0.7895,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753088185796712e-06,
+      "loss": 0.7907,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747013484539127e-06,
+      "loss": 0.7605,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740939659010292e-06,
+      "loss": 0.6975,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734866711571083e-06,
+      "loss": 0.7549,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728794644582035e-06,
+      "loss": 0.7084,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722723460403342e-06,
+      "loss": 0.3391,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716653161394852e-06,
+      "loss": 0.8313,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710583749916072e-06,
+      "loss": 0.7553,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704515228326169e-06,
+      "loss": 0.7519,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698447598983946e-06,
+      "loss": 0.7795,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692380864247882e-06,
+      "loss": 0.8099,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68631502647609e-06,
+      "loss": 0.7625,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.680250088026345e-06,
+      "loss": 0.7627,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674186051256071e-06,
+      "loss": 0.7413,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668122918522342e-06,
+      "loss": 0.8014,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662060692181873e-06,
+      "loss": 0.8001,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655999374591034e-06,
+      "loss": 0.8155,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649938968105846e-06,
+      "loss": 0.7956,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643879475081961e-06,
+      "loss": 0.8024,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637820897874688e-06,
+      "loss": 0.7834,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631763238838979e-06,
+      "loss": 0.7695,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625706500329424e-06,
+      "loss": 0.7233,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6196506847002586e-06,
+      "loss": 0.343,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61359579430536e-06,
+      "loss": 0.7826,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607541831498247e-06,
+      "loss": 0.7377,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601488798632075e-06,
+      "loss": 0.76,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595436698059637e-06,
+      "loss": 0.7597,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589385532133365e-06,
+      "loss": 0.7831,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583335303205328e-06,
+      "loss": 0.6889,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5772860136272315e-06,
+      "loss": 0.8813,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57123766575042e-06,
+      "loss": 0.8012,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565190261925859e-06,
+      "loss": 0.8101,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55914380450416e-06,
+      "loss": 0.832,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553098295835562e-06,
+      "loss": 0.6861,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547053738269937e-06,
+      "loss": 0.8142,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54101013415678e-06,
+      "loss": 0.8411,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534967485845225e-06,
+      "loss": 0.7646,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528925795684028e-06,
+      "loss": 0.8036,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522885066021575e-06,
+      "loss": 0.837,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51684529920588e-06,
+      "loss": 0.7676,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510806497584582e-06,
+      "loss": 0.7469,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504768663504942e-06,
+      "loss": 0.7641,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498731799313852e-06,
+      "loss": 0.7697,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492695907357812e-06,
+      "loss": 0.8256,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486660989982962e-06,
+      "loss": 0.7874,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480627049535054e-06,
+      "loss": 0.7713,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474594088359459e-06,
+      "loss": 0.7776,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468562108801176e-06,
+      "loss": 0.7301,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46253111320481e-06,
+      "loss": 0.7865,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456501103914593e-06,
+      "loss": 0.7335,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450472083274369e-06,
+      "loss": 0.3444,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4444440536276055e-06,
+      "loss": 0.7743,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438417017317369e-06,
+      "loss": 0.7704,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4323909766863555e-06,
+      "loss": 0.8211,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426365934076865e-06,
+      "loss": 0.8217,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4203418918308135e-06,
+      "loss": 0.7722,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414318852289728e-06,
+      "loss": 0.7512,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408296817794747e-06,
+      "loss": 0.7484,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402275790686611e-06,
+      "loss": 0.8135,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.396255773305681e-06,
+      "loss": 0.783,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390236767991912e-06,
+      "loss": 0.7414,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384218777084875e-06,
+      "loss": 0.7749,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378201802923743e-06,
+      "loss": 0.7606,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372185847847301e-06,
+      "loss": 0.8328,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366170914193923e-06,
+      "loss": 0.7237,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360157004301599e-06,
+      "loss": 0.7943,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354144120507918e-06,
+      "loss": 0.7453,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34813226515007e-06,
+      "loss": 0.8098,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342121440564837e-06,
+      "loss": 0.7747,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336111649088615e-06,
+      "loss": 0.7453,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330102893057387e-06,
+      "loss": 0.7612,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324095174806738e-06,
+      "loss": 0.776,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318088496671851e-06,
+      "loss": 0.7561,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312082860987501e-06,
+      "loss": 0.7651,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306078270088061e-06,
+      "loss": 0.8005,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300074726307496e-06,
+      "loss": 0.7119,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2940722319793665e-06,
+      "loss": 0.8366,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288070789436819e-06,
+      "loss": 0.738,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282070401012595e-06,
+      "loss": 0.799,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276071069039033e-06,
+      "loss": 0.798,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.270072795848047e-06,
+      "loss": 0.3455,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264075583771149e-06,
+      "loss": 0.8012,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258079435139439e-06,
+      "loss": 0.8151,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2520843522836014e-06,
+      "loss": 0.7731,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246090337533904e-06,
+      "loss": 0.7741,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240097393220201e-06,
+      "loss": 0.793,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23410552167193e-06,
+      "loss": 0.7814,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228114725218114e-06,
+      "loss": 0.8748,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222125006187357e-06,
+      "loss": 0.7794,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216136366907844e-06,
+      "loss": 0.8082,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21014880970734e-06,
+      "loss": 0.8145,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204162336913187e-06,
+      "loss": 0.8345,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198176950852314e-06,
+      "loss": 0.3143,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192192653851219e-06,
+      "loss": 0.7264,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186209448235974e-06,
+      "loss": 0.7935,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18022733633224e-06,
+      "loss": 0.7315,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1742463204652415e-06,
+      "loss": 0.7751,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1682664029597785e-06,
+      "loss": 0.8411,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1622875861402285e-06,
+      "loss": 0.767,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156309872330538e-06,
+      "loss": 0.7654,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1503332638542285e-06,
+      "loss": 0.7407,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144357763034386e-06,
+      "loss": 0.832,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13838337219367e-06,
+      "loss": 0.8191,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132410093654304e-06,
+      "loss": 0.8123,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.126437929738085e-06,
+      "loss": 0.811,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120466882766374e-06,
+      "loss": 0.8075,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1144969550601e-06,
+      "loss": 0.8369,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1085281489397515e-06,
+      "loss": 0.7081,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102560466725387e-06,
+      "loss": 0.772,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096593910736624e-06,
+      "loss": 0.8022,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090628483292652e-06,
+      "loss": 0.7984,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0846641867122e-06,
+      "loss": 0.7793,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078701023313582e-06,
+      "loss": 0.8217,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072738995414654e-06,
+      "loss": 0.769,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066778105332842e-06,
+      "loss": 0.3186,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060818355385124e-06,
+      "loss": 0.8628,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054859747888037e-06,
+      "loss": 0.7616,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0489022851576685e-06,
+      "loss": 0.866,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042945969509673e-06,
+      "loss": 0.8099,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.036990803259245e-06,
+      "loss": 0.7631,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0310367887211394e-06,
+      "loss": 0.776,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0250839282096635e-06,
+      "loss": 0.699,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019132224038681e-06,
+      "loss": 0.7044,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013181678521593e-06,
+      "loss": 0.8119,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0072322939713585e-06,
+      "loss": 0.7734,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0012840727004895e-06,
+      "loss": 0.8201,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9953370170210374e-06,
+      "loss": 0.7797,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.989391129244607e-06,
+      "loss": 0.728,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983446411682343e-06,
+      "loss": 0.7648,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9775028666449365e-06,
+      "loss": 0.7931,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971560496442627e-06,
+      "loss": 0.7732,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965619303385196e-06,
+      "loss": 0.7887,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959679289781968e-06,
+      "loss": 0.845,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953740457941803e-06,
+      "loss": 0.8051,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9478028101731086e-06,
+      "loss": 0.8544,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941866348783835e-06,
+      "loss": 0.7512,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9359310760814545e-06,
+      "loss": 0.8506,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929996994372995e-06,
+      "loss": 0.7944,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924064105965018e-06,
+      "loss": 0.8342,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918132413163614e-06,
+      "loss": 0.7983,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912201918274415e-06,
+      "loss": 0.7534,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906272623602587e-06,
+      "loss": 0.3416,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900344531452825e-06,
+      "loss": 0.7749,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894417644129368e-06,
+      "loss": 0.745,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888491963935968e-06,
+      "loss": 0.811,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882567493175921e-06,
+      "loss": 0.7888,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876644234152051e-06,
+      "loss": 0.7939,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8707221891667105e-06,
+      "loss": 0.7881,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86480136052178e-06,
+      "loss": 0.722,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8588817505186685e-06,
+      "loss": 0.7974,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852963361458305e-06,
+      "loss": 0.7999,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847046195641151e-06,
+      "loss": 0.3327,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841130255367196e-06,
+      "loss": 0.7725,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835215542935938e-06,
+      "loss": 0.7465,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82930206064641e-06,
+      "loss": 0.7747,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823389810797167e-06,
+      "loss": 0.7902,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.817478795686279e-06,
+      "loss": 0.3414,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811569017611337e-06,
+      "loss": 0.7366,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805660478869458e-06,
+      "loss": 0.8527,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799753181757271e-06,
+      "loss": 0.8498,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793847128570923e-06,
+      "loss": 0.3325,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787942321606078e-06,
+      "loss": 0.6936,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782038763157915e-06,
+      "loss": 0.8196,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776136455521128e-06,
+      "loss": 0.8258,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770235400989926e-06,
+      "loss": 0.7388,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764335601858034e-06,
+      "loss": 0.7695,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758437060418681e-06,
+      "loss": 0.7693,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752539778964613e-06,
+      "loss": 0.8019,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746643759788083e-06,
+      "loss": 0.6712,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740749005180862e-06,
+      "loss": 0.8036,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734855517434212e-06,
+      "loss": 0.7545,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728963298838921e-06,
+      "loss": 0.803,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72307235168527e-06,
+      "loss": 0.7843,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.717182678263055e-06,
+      "loss": 0.7188,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711294280861573e-06,
+      "loss": 0.8211,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705407161769628e-06,
+      "loss": 0.7574,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69952132327552e-06,
+      "loss": 0.7452,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69363676766706e-06,
+      "loss": 0.7595,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687753497231552e-06,
+      "loss": 0.8432,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681871514255807e-06,
+      "loss": 0.8244,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675990821026131e-06,
+      "loss": 0.7251,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6701114198283355e-06,
+      "loss": 0.7575,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664233312947721e-06,
+      "loss": 0.778,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658356502669091e-06,
+      "loss": 0.8131,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652480991276742e-06,
+      "loss": 0.7441,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6466067810544676e-06,
+      "loss": 0.8074,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64073387428556e-06,
+      "loss": 0.7384,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6348622732527915e-06,
+      "loss": 0.7666,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628991980238435e-06,
+      "loss": 0.813,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623122997524257e-06,
+      "loss": 0.773,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617255327391514e-06,
+      "loss": 0.7505,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611388972120949e-06,
+      "loss": 0.8334,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6055239339928e-06,
+      "loss": 0.3241,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599660215286783e-06,
+      "loss": 0.809,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593797818282113e-06,
+      "loss": 0.7269,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5879367452574775e-06,
+      "loss": 0.7953,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582076998491061e-06,
+      "loss": 0.7745,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5762185802605294e-06,
+      "loss": 0.7503,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5703614928430336e-06,
+      "loss": 0.8083,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564505738515199e-06,
+      "loss": 0.8408,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558651319553142e-06,
+      "loss": 0.7864,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552798238232457e-06,
+      "loss": 0.7448,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54694649682822e-06,
+      "loss": 0.7939,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541096097614983e-06,
+      "loss": 0.7848,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535247042866777e-06,
+      "loss": 0.3259,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529399334857109e-06,
+      "loss": 0.7583,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523552975858967e-06,
+      "loss": 0.8543,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517707968144812e-06,
+      "loss": 0.6901,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.511864313986582e-06,
+      "loss": 0.7124,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506022015655684e-06,
+      "loss": 0.8194,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500181075423001e-06,
+      "loss": 0.8324,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494341495558894e-06,
+      "loss": 0.718,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4885032783331804e-06,
+      "loss": 0.7819,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482666426015161e-06,
+      "loss": 0.7936,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.476830940873603e-06,
+      "loss": 0.8189,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470996825176739e-06,
+      "loss": 0.8166,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46516408119227e-06,
+      "loss": 0.7478,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4593327111873695e-06,
+      "loss": 0.7449,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45350271742867e-06,
+      "loss": 0.7438,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447674102182275e-06,
+      "loss": 0.8112,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441846867713744e-06,
+      "loss": 0.7745,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4360210162881054e-06,
+      "loss": 0.796,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430196550169849e-06,
+      "loss": 0.7582,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42437347162293e-06,
+      "loss": 0.7561,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41855178291076e-06,
+      "loss": 0.7681,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.412731486296206e-06,
+      "loss": 0.8245,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406912584041605e-06,
+      "loss": 0.7499,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401095078408741e-06,
+      "loss": 0.3337,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395278971658867e-06,
+      "loss": 0.7886,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389464266052679e-06,
+      "loss": 0.7676,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383650963850336e-06,
+      "loss": 0.7554,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377839067311449e-06,
+      "loss": 0.7924,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3720285786950865e-06,
+      "loss": 0.655,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366219500259766e-06,
+      "loss": 0.7708,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360411834263458e-06,
+      "loss": 0.7916,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354605582963587e-06,
+      "loss": 0.7972,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348800748617022e-06,
+      "loss": 0.7623,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342997333480081e-06,
+      "loss": 0.7035,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337195339808535e-06,
+      "loss": 0.7656,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3313947698576006e-06,
+      "loss": 0.782,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3255956258819415e-06,
+      "loss": 0.7375,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3197979101356696e-06,
+      "loss": 0.76,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314001624872335e-06,
+      "loss": 0.7646,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308206772344935e-06,
+      "loss": 0.7701,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302413354805912e-06,
+      "loss": 0.7984,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296621374507154e-06,
+      "loss": 0.7899,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290830833699975e-06,
+      "loss": 0.7833,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285041734635149e-06,
+      "loss": 0.7911,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279254079562873e-06,
+      "loss": 0.7931,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273467870732797e-06,
+      "loss": 0.8149,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267683110393996e-06,
+      "loss": 0.8281,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261899800794996e-06,
+      "loss": 0.7314,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2561179441837435e-06,
+      "loss": 0.7626,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250337542807636e-06,
+      "loss": 0.8156,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244558598913487e-06,
+      "loss": 0.7496,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238781114747561e-06,
+      "loss": 0.7723,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233005092555544e-06,
+      "loss": 0.8038,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2272305345825624e-06,
+      "loss": 0.7564,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221457443073164e-06,
+      "loss": 0.7637,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215685820271332e-06,
+      "loss": 0.8246,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.20991566842048e-06,
+      "loss": 0.7712,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204146989763449e-06,
+      "loss": 0.7902,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198379786542503e-06,
+      "loss": 0.7977,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192614060999337e-06,
+      "loss": 0.776,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1868498153750685e-06,
+      "loss": 0.7705,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181087051910242e-06,
+      "loss": 0.7771,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175325772844828e-06,
+      "loss": 0.8025,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1695659804182175e-06,
+      "loss": 0.7181,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16380767686922e-06,
+      "loss": 0.7969,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158050864436073e-06,
+      "loss": 0.7558,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152295545356434e-06,
+      "loss": 0.7795,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14654172186737e-06,
+      "loss": 0.7159,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1407893962053765e-06,
+      "loss": 0.7545,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1350385706063685e-06,
+      "loss": 0.7485,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12928924730567e-06,
+      "loss": 0.786,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123541428538025e-06,
+      "loss": 0.7327,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117795116537593e-06,
+      "loss": 0.7618,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11205031353795e-06,
+      "loss": 0.8365,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.10630702177208e-06,
+      "loss": 0.8088,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1005652434723825e-06,
+      "loss": 0.7644,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094824980870669e-06,
+      "loss": 0.8275,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089086236198159e-06,
+      "loss": 0.7585,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083349011685487e-06,
+      "loss": 0.802,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077613309562697e-06,
+      "loss": 0.8208,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0718791320592325e-06,
+      "loss": 0.7894,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066146481403953e-06,
+      "loss": 0.7312,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060415359825123e-06,
+      "loss": 0.7425,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0546857695504115e-06,
+      "loss": 0.7545,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048957712806888e-06,
+      "loss": 0.7969,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043231191821033e-06,
+      "loss": 0.7472,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037506208818728e-06,
+      "loss": 0.7569,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031782766025253e-06,
+      "loss": 0.7899,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026060865665293e-06,
+      "loss": 0.7509,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020340509962935e-06,
+      "loss": 0.834,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014621701141663e-06,
+      "loss": 0.7273,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008904441424359e-06,
+      "loss": 0.771,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003188733033303e-06,
+      "loss": 0.321,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.997474578190171e-06,
+      "loss": 0.7895,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9917619791160395e-06,
+      "loss": 0.7507,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986050938031378e-06,
+      "loss": 0.7595,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980341457156051e-06,
+      "loss": 0.8262,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974633538709311e-06,
+      "loss": 0.8062,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.968927184909812e-06,
+      "loss": 0.7616,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963222397975595e-06,
+      "loss": 0.7818,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957519180124097e-06,
+      "loss": 0.8023,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951817533572133e-06,
+      "loss": 0.7221,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9461174605359185e-06,
+      "loss": 0.7725,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.940418963231051e-06,
+      "loss": 0.7856,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934722043872524e-06,
+      "loss": 0.7907,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929026704674707e-06,
+      "loss": 0.7791,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923332947851365e-06,
+      "loss": 0.7985,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917640775615641e-06,
+      "loss": 0.8629,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.911950190180067e-06,
+      "loss": 0.7973,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9062611937565485e-06,
+      "loss": 0.8183,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900573788556384e-06,
+      "loss": 0.8098,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894887976790251e-06,
+      "loss": 0.7854,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8892037606682094e-06,
+      "loss": 0.7597,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883521142399691e-06,
+      "loss": 0.808,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877840124193513e-06,
+      "loss": 0.8255,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87216070825787e-06,
+      "loss": 0.789,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866482896800336e-06,
+      "loss": 0.7852,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860806692027854e-06,
+      "loss": 0.7584,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85513209614675e-06,
+      "loss": 0.7471,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8494591113627165e-06,
+      "loss": 0.802,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843787739880827e-06,
+      "loss": 0.3702,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8381179839055305e-06,
+      "loss": 0.7623,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8324498456406344e-06,
+      "loss": 0.7009,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826783327289336e-06,
+      "loss": 0.7411,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8211184310541835e-06,
+      "loss": 0.7307,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815455159137118e-06,
+      "loss": 0.751,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809793513739421e-06,
+      "loss": 0.83,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804133497061759e-06,
+      "loss": 0.85,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798475111304168e-06,
+      "loss": 0.7767,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7928183586660395e-06,
+      "loss": 0.8275,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787163241346141e-06,
+      "loss": 0.3233,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.781509761542594e-06,
+      "loss": 0.8174,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7758579214528875e-06,
+      "loss": 0.7787,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770207723273881e-06,
+      "loss": 0.7345,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764559169201783e-06,
+      "loss": 0.7459,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75891226143217e-06,
+      "loss": 0.7876,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753267002159972e-06,
+      "loss": 0.8041,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747623393579494e-06,
+      "loss": 0.7429,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741981437884381e-06,
+      "loss": 0.7953,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736341137267642e-06,
+      "loss": 0.817,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730702493921653e-06,
+      "loss": 0.7672,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72506551003813e-06,
+      "loss": 0.7424,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719430187808153e-06,
+      "loss": 0.6994,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7137965294221534e-06,
+      "loss": 0.7631,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708164537069911e-06,
+      "loss": 0.8296,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.702534212940575e-06,
+      "loss": 0.794,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.696905559222623e-06,
+      "loss": 0.741,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691278578103905e-06,
+      "loss": 0.7901,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685653271771605e-06,
+      "loss": 0.7488,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680029642412259e-06,
+      "loss": 0.3183,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6744076922117665e-06,
+      "loss": 0.7929,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668787423355344e-06,
+      "loss": 0.7729,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663168838027587e-06,
+      "loss": 0.7762,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.657551938412416e-06,
+      "loss": 0.7702,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651936726693099e-06,
+      "loss": 0.7922,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6463232050522575e-06,
+      "loss": 0.7595,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640711375671843e-06,
+      "loss": 0.8306,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635101240733164e-06,
+      "loss": 0.8066,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629492802416858e-06,
+      "loss": 0.7973,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623886062902906e-06,
+      "loss": 0.7362,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618281024370632e-06,
+      "loss": 0.8019,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612677688998691e-06,
+      "loss": 0.77,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607076058965092e-06,
+      "loss": 0.7184,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601476136447165e-06,
+      "loss": 0.7431,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595877923621579e-06,
+      "loss": 0.351,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5902814226643485e-06,
+      "loss": 0.7882,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584686635750809e-06,
+      "loss": 0.7465,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579093565055649e-06,
+      "loss": 0.802,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573502212752861e-06,
+      "loss": 0.8024,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567912581015797e-06,
+      "loss": 0.7473,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562324672017125e-06,
+      "loss": 0.7906,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556738487928845e-06,
+      "loss": 0.8054,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551154030922297e-06,
+      "loss": 0.7813,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5455713031681366e-06,
+      "loss": 0.8146,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539990306836349e-06,
+      "loss": 0.8509,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534411044096259e-06,
+      "loss": 0.7068,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528833517116504e-06,
+      "loss": 0.7487,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5232577280650505e-06,
+      "loss": 0.7353,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5176836791091846e-06,
+      "loss": 0.8006,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512111372415533e-06,
+      "loss": 0.7957,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5065408101500275e-06,
+      "loss": 0.7414,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500971994477924e-06,
+      "loss": 0.7352,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.495404927563812e-06,
+      "loss": 0.7117,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489839611571586e-06,
+      "loss": 0.7377,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484276048664477e-06,
+      "loss": 0.751,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4787142410050086e-06,
+      "loss": 0.7491,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.47315419075505e-06,
+      "loss": 0.7983,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.467595900075773e-06,
+      "loss": 0.8416,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.462039371127661e-06,
+      "loss": 0.7414,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456484606070529e-06,
+      "loss": 0.7453,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4509316070634925e-06,
+      "loss": 0.7798,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445380376264979e-06,
+      "loss": 0.781,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439830915832745e-06,
+      "loss": 0.7677,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434283227923842e-06,
+      "loss": 0.7268,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428737314694639e-06,
+      "loss": 0.7444,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423193178300813e-06,
+      "loss": 0.8107,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4176508208973574e-06,
+      "loss": 0.8135,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4121102446385685e-06,
+      "loss": 0.7849,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406571451678045e-06,
+      "loss": 0.8005,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401034444168705e-06,
+      "loss": 0.7414,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395499224262763e-06,
+      "loss": 0.7842,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3899657941117426e-06,
+      "loss": 0.7657,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384434155866467e-06,
+      "loss": 0.3291,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378904311677067e-06,
+      "loss": 0.7963,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373376263692982e-06,
+      "loss": 0.6836,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367850014062937e-06,
+      "loss": 0.7951,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362325564934978e-06,
+      "loss": 0.7831,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356802918456437e-06,
+      "loss": 0.761,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351282076773945e-06,
+      "loss": 0.3441,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.34576304203344e-06,
+      "loss": 0.7713,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340245816380156e-06,
+      "loss": 0.7663,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334730401958616e-06,
+      "loss": 0.7882,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.329216800912646e-06,
+      "loss": 0.7955,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323705015385362e-06,
+      "loss": 0.8044,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318195047519184e-06,
+      "loss": 0.8157,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312686899455813e-06,
+      "loss": 0.7064,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3071805733362555e-06,
+      "loss": 0.768,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3016760713008e-06,
+      "loss": 0.8258,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296173395489028e-06,
+      "loss": 0.8005,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290672548039812e-06,
+      "loss": 0.7123,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285173531091312e-06,
+      "loss": 0.8197,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279676346780984e-06,
+      "loss": 0.7448,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274180997245564e-06,
+      "loss": 0.7252,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268687484621073e-06,
+      "loss": 0.8349,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263195811042829e-06,
+      "loss": 0.8191,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2577059786454196e-06,
+      "loss": 0.749,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25221798956274e-06,
+      "loss": 0.7719,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246731845927937e-06,
+      "loss": 0.8224,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2412475498734675e-06,
+      "loss": 0.7984,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235765103531058e-06,
+      "loss": 0.365,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230284509031716e-06,
+      "loss": 0.7484,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224805768505734e-06,
+      "loss": 0.8231,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2193288840826775e-06,
+      "loss": 0.7594,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2138538578914e-06,
+      "loss": 0.7466,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208380692060023e-06,
+      "loss": 0.7518,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202909388715951e-06,
+      "loss": 0.3092,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1974399499858595e-06,
+      "loss": 0.7005,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1919723779957e-06,
+      "loss": 0.7892,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186506674870708e-06,
+      "loss": 0.7888,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181042842735381e-06,
+      "loss": 0.766,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175580883713489e-06,
+      "loss": 0.3417,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170120799928087e-06,
+      "loss": 0.755,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164662593501484e-06,
+      "loss": 0.7178,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159206266555278e-06,
+      "loss": 0.7265,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1537518212103125e-06,
+      "loss": 0.7481,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148299259586723e-06,
+      "loss": 0.7704,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142848583803901e-06,
+      "loss": 0.7505,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137399795980502e-06,
+      "loss": 0.7599,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1319528982344614e-06,
+      "loss": 0.8091,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126507892682966e-06,
+      "loss": 0.745,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121064781442469e-06,
+      "loss": 0.862,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115623566628698e-06,
+      "loss": 0.7943,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.110184250356633e-06,
+      "loss": 0.7492,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104746834740518e-06,
+      "loss": 0.7762,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099311321893856e-06,
+      "loss": 0.738,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09387771392942e-06,
+      "loss": 0.7921,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088446012959236e-06,
+      "loss": 0.7442,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083016221094582e-06,
+      "loss": 0.7368,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077588340446009e-06,
+      "loss": 0.82,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0721623731233145e-06,
+      "loss": 0.8071,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066738321235554e-06,
+      "loss": 0.8322,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061316186891039e-06,
+      "loss": 0.7468,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055895972197334e-06,
+      "loss": 0.7623,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050477679261264e-06,
+      "loss": 0.8248,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.045061310188897e-06,
+      "loss": 0.7989,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0396468670855654e-06,
+      "loss": 0.7893,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034234352055843e-06,
+      "loss": 0.7946,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0288237672035526e-06,
+      "loss": 0.801,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02341511463178e-06,
+      "loss": 0.8297,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018008396442847e-06,
+      "loss": 0.7332,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012603614738328e-06,
+      "loss": 0.6961,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007200771619045e-06,
+      "loss": 0.7913,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001799869185061e-06,
+      "loss": 0.7927,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996400909535698e-06,
+      "loss": 0.7512,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991003894769508e-06,
+      "loss": 0.7242,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9856088269842995e-06,
+      "loss": 0.7751,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980215708277116e-06,
+      "loss": 0.3271,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974824540744245e-06,
+      "loss": 0.7274,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969435326481217e-06,
+      "loss": 0.7959,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964048067582797e-06,
+      "loss": 0.7615,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958662766143006e-06,
+      "loss": 0.7449,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953279424255083e-06,
+      "loss": 0.7565,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947898044011525e-06,
+      "loss": 0.7726,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942518627504054e-06,
+      "loss": 0.766,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9371411768236286e-06,
+      "loss": 0.7745,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931765694060457e-06,
+      "loss": 0.7337,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92639218130396e-06,
+      "loss": 0.7772,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921020640642812e-06,
+      "loss": 0.7415,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9156510741649135e-06,
+      "loss": 0.7453,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9102834839573934e-06,
+      "loss": 0.7974,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904917872106625e-06,
+      "loss": 0.811,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899554240698195e-06,
+      "loss": 0.729,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894192591816942e-06,
+      "loss": 0.3452,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888832927546914e-06,
+      "loss": 0.7786,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.883475249971396e-06,
+      "loss": 0.808,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878119561172904e-06,
+      "loss": 0.8031,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.87276586323317e-06,
+      "loss": 0.7999,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867414158233167e-06,
+      "loss": 0.7362,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862064448253084e-06,
+      "loss": 0.7712,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856716735372332e-06,
+      "loss": 0.8087,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851371021669556e-06,
+      "loss": 0.7536,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846027309222614e-06,
+      "loss": 0.8219,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840685600108598e-06,
+      "loss": 0.8011,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835345896403801e-06,
+      "loss": 0.7685,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8300082001837575e-06,
+      "loss": 0.7819,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.824672513523212e-06,
+      "loss": 0.334,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8193388384961235e-06,
+      "loss": 0.7238,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814007177175683e-06,
+      "loss": 0.791,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808677531634288e-06,
+      "loss": 0.814,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803349903943549e-06,
+      "loss": 0.7836,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798024296174306e-06,
+      "loss": 0.7981,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792700710396603e-06,
+      "loss": 0.3424,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787379148679699e-06,
+      "loss": 0.7408,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7820596130920685e-06,
+      "loss": 0.7823,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776742105701402e-06,
+      "loss": 0.742,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771426628574594e-06,
+      "loss": 0.8353,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.766113183777753e-06,
+      "loss": 0.7145,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7608017733762035e-06,
+      "loss": 0.7542,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7554923994344674e-06,
+      "loss": 0.7464,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750185064016292e-06,
+      "loss": 0.7926,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.744879769184611e-06,
+      "loss": 0.3209,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739576517001575e-06,
+      "loss": 0.8004,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.73427530952855e-06,
+      "loss": 0.7427,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728976148826091e-06,
+      "loss": 0.7989,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723679036953972e-06,
+      "loss": 0.7343,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71838397597116e-06,
+      "loss": 0.7143,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713090967935825e-06,
+      "loss": 0.7339,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707800014905351e-06,
+      "loss": 0.7073,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702511118936308e-06,
+      "loss": 0.7649,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6972242820844774e-06,
+      "loss": 0.3119,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691939506404828e-06,
+      "loss": 0.7064,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686656793951545e-06,
+      "loss": 0.7138,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681376146777999e-06,
+      "loss": 0.7794,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.676097566936756e-06,
+      "loss": 0.7657,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670821056479592e-06,
+      "loss": 0.8403,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6655466174574635e-06,
+      "loss": 0.7161,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6602742519205325e-06,
+      "loss": 0.7655,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655003961918146e-06,
+      "loss": 0.7947,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64973574949885e-06,
+      "loss": 0.7599,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644469616710386e-06,
+      "loss": 0.7771,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639205565599678e-06,
+      "loss": 0.7966,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633943598212852e-06,
+      "loss": 0.7555,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628683716595216e-06,
+      "loss": 0.7817,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623425922791264e-06,
+      "loss": 0.8272,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618170218844692e-06,
+      "loss": 0.7057,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612916606798371e-06,
+      "loss": 0.7819,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607665088694365e-06,
+      "loss": 0.742,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.602415666573922e-06,
+      "loss": 0.7174,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597168342477472e-06,
+      "loss": 0.7585,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59192311844464e-06,
+      "loss": 0.7255,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5866799965142205e-06,
+      "loss": 0.8566,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581438978724208e-06,
+      "loss": 0.3327,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576200067111763e-06,
+      "loss": 0.7701,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570963263713236e-06,
+      "loss": 0.7648,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565728570564156e-06,
+      "loss": 0.7426,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560495989699226e-06,
+      "loss": 0.7584,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555265523152344e-06,
+      "loss": 0.7231,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55003717295657e-06,
+      "loss": 0.6802,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544810941144143e-06,
+      "loss": 0.816,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539586829746492e-06,
+      "loss": 0.8102,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534364840794204e-06,
+      "loss": 0.7967,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5291449763170614e-06,
+      "loss": 0.7133,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5239272383439936e-06,
+      "loss": 0.7723,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518711628903131e-06,
+      "loss": 0.7785,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51349815002176e-06,
+      "loss": 0.8355,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.50828680372634e-06,
+      "loss": 0.75,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503077592042512e-06,
+      "loss": 0.7814,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497870516995073e-06,
+      "loss": 0.8139,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.492665580608005e-06,
+      "loss": 0.7796,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4874627849044435e-06,
+      "loss": 0.7734,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482262131906702e-06,
+      "loss": 0.8407,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477063623636258e-06,
+      "loss": 0.7438,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471867262113749e-06,
+      "loss": 0.7891,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466673049358993e-06,
+      "loss": 0.7737,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46148098739096e-06,
+      "loss": 0.7771,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456291078227787e-06,
+      "loss": 0.8079,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451103323886778e-06,
+      "loss": 0.7457,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445917726384395e-06,
+      "loss": 0.7475,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440734287736268e-06,
+      "loss": 0.79,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435553009957174e-06,
+      "loss": 0.7662,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4303738950610675e-06,
+      "loss": 0.6733,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425196945061052e-06,
+      "loss": 0.8126,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420022161969386e-06,
+      "loss": 0.7637,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414849547797501e-06,
+      "loss": 0.8094,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40967910455597e-06,
+      "loss": 0.7978,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404510834254526e-06,
+      "loss": 0.7172,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399344738902066e-06,
+      "loss": 0.7385,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39418082050663e-06,
+      "loss": 0.8363,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38901908107542e-06,
+      "loss": 0.7932,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383859522614781e-06,
+      "loss": 0.7869,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378702147130226e-06,
+      "loss": 0.8095,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3735469566264065e-06,
+      "loss": 0.746,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368393953107125e-06,
+      "loss": 0.8669,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363243138575345e-06,
+      "loss": 0.7457,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3580945150331685e-06,
+      "loss": 0.8062,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35294808448185e-06,
+      "loss": 0.8003,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347803848921789e-06,
+      "loss": 0.7955,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342661810352531e-06,
+      "loss": 0.6969,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337521970772778e-06,
+      "loss": 0.777,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332384332180362e-06,
+      "loss": 0.7831,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327248896572272e-06,
+      "loss": 0.7865,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322115665944634e-06,
+      "loss": 0.7983,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3169846422927154e-06,
+      "loss": 0.7443,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3118558276109345e-06,
+      "loss": 0.7338,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3067292238928435e-06,
+      "loss": 0.7504,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301604833131136e-06,
+      "loss": 0.7313,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296482657317646e-06,
+      "loss": 0.7327,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291362698443344e-06,
+      "loss": 0.7941,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28624495849835e-06,
+      "loss": 0.3621,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281129439471906e-06,
+      "loss": 0.3226,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276016143352404e-06,
+      "loss": 0.7236,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270905072127362e-06,
+      "loss": 0.7713,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265796227783441e-06,
+      "loss": 0.7638,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2606896123064274e-06,
+      "loss": 0.7866,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255585227681247e-06,
+      "loss": 0.7047,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2504830758919646e-06,
+      "loss": 0.8244,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245383158921764e-06,
+      "loss": 0.7603,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2402854787529665e-06,
+      "loss": 0.7871,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235190037367031e-06,
+      "loss": 0.8471,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2300968367445304e-06,
+      "loss": 0.8095,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.225005878865185e-06,
+      "loss": 0.7701,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219917165707829e-06,
+      "loss": 0.7266,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214830699250431e-06,
+      "loss": 0.6842,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209746481470081e-06,
+      "loss": 0.7223,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204664514342999e-06,
+      "loss": 0.8065,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199584799844534e-06,
+      "loss": 0.7662,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1945073399491486e-06,
+      "loss": 0.7683,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189432136630442e-06,
+      "loss": 0.3278,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184359191861128e-06,
+      "loss": 0.7572,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179288507613042e-06,
+      "loss": 0.8311,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1742200858571445e-06,
+      "loss": 0.718,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169153928563511e-06,
+      "loss": 0.8349,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164090037701347e-06,
+      "loss": 0.8029,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159028415238968e-06,
+      "loss": 0.77,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153969063143808e-06,
+      "loss": 0.707,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148911983382428e-06,
+      "loss": 0.7338,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143857177920491e-06,
+      "loss": 0.7811,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138804648722795e-06,
+      "loss": 0.7338,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133754397753229e-06,
+      "loss": 0.7753,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128706426974818e-06,
+      "loss": 0.7342,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1236607383496904e-06,
+      "loss": 0.8222,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118617333839085e-06,
+      "loss": 0.7228,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113576215403366e-06,
+      "loss": 0.7219,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.108537385001995e-06,
+      "loss": 0.7895,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103500844593546e-06,
+      "loss": 0.6948,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098466596135715e-06,
+      "loss": 0.8067,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093434641585293e-06,
+      "loss": 0.7879,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088404982898186e-06,
+      "loss": 0.314,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083377622029403e-06,
+      "loss": 0.7577,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078352560933069e-06,
+      "loss": 0.779,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0733298015624075e-06,
+      "loss": 0.7736,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068309345869745e-06,
+      "loss": 0.7284,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063291195806524e-06,
+      "loss": 0.7794,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05827535332328e-06,
+      "loss": 0.7862,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053261820369655e-06,
+      "loss": 0.7294,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0482505988943935e-06,
+      "loss": 0.7855,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043241690845338e-06,
+      "loss": 0.7826,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038235098169442e-06,
+      "loss": 0.7817,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033230822812745e-06,
+      "loss": 0.7551,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028228866720402e-06,
+      "loss": 0.8273,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023229231836651e-06,
+      "loss": 0.825,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0182319201048335e-06,
+      "loss": 0.7267,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0132369334673945e-06,
+      "loss": 0.7535,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.008244273865867e-06,
+      "loss": 0.7415,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00325394324088e-06,
+      "loss": 0.8301,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998265943532161e-06,
+      "loss": 0.7412,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993280276678526e-06,
+      "loss": 0.3543,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988296944617894e-06,
+      "loss": 0.8029,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983315949287265e-06,
+      "loss": 0.8117,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978337292622743e-06,
+      "loss": 0.7429,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973360976559512e-06,
+      "loss": 0.7472,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96838700303185e-06,
+      "loss": 0.7555,
+      "step": 3714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963415373973124e-06,
+      "loss": 0.6487,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958446091315789e-06,
+      "loss": 0.7604,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953479156991396e-06,
+      "loss": 0.3494,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9485145729305685e-06,
+      "loss": 0.7592,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943552341063034e-06,
+      "loss": 0.7308,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93859246331759e-06,
+      "loss": 0.7213,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933634941622124e-06,
+      "loss": 0.7853,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928679777903616e-06,
+      "loss": 0.6797,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923726974088117e-06,
+      "loss": 0.7925,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918776532100768e-06,
+      "loss": 0.7969,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913828453865789e-06,
+      "loss": 0.7747,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90888274130648e-06,
+      "loss": 0.7877,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903939396345231e-06,
+      "loss": 0.7468,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898998420903497e-06,
+      "loss": 0.8117,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894059816901826e-06,
+      "loss": 0.3337,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889123586259835e-06,
+      "loss": 0.8683,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884189730896221e-06,
+      "loss": 0.739,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879258252728758e-06,
+      "loss": 0.8166,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874329153674293e-06,
+      "loss": 0.3144,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869402435648758e-06,
+      "loss": 0.7321,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864478100567149e-06,
+      "loss": 0.7983,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859556150343536e-06,
+      "loss": 0.758,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854636586891073e-06,
+      "loss": 0.773,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849719412121971e-06,
+      "loss": 0.8144,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844804627947528e-06,
+      "loss": 0.7295,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839892236278102e-06,
+      "loss": 0.8053,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8349822390231225e-06,
+      "loss": 0.7903,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830074638091092e-06,
+      "loss": 0.7851,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825169435389574e-06,
+      "loss": 0.766,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820266632825214e-06,
+      "loss": 0.8173,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815366232303711e-06,
+      "loss": 0.7442,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8104682357298327e-06,
+      "loss": 0.7404,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8055726450074215e-06,
+      "loss": 0.7296,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800679462039375e-06,
+      "loss": 0.7615,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7957886887276574e-06,
+      "loss": 0.3603,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7909003269732926e-06,
+      "loss": 0.3527,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7860143786763805e-06,
+      "loss": 0.7593,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7811308457360683e-06,
+      "loss": 0.7522,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7762497300505676e-06,
+      "loss": 0.8216,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771371033517159e-06,
+      "loss": 0.7732,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.766494758032171e-06,
+      "loss": 0.7368,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7616209054910047e-06,
+      "loss": 0.8056,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756749477788101e-06,
+      "loss": 0.7477,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7518804768169704e-06,
+      "loss": 0.7525,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7470139044701824e-06,
+      "loss": 0.7846,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7421497626393545e-06,
+      "loss": 0.7775,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737288053215168e-06,
+      "loss": 0.7619,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7324287780873503e-06,
+      "loss": 0.3044,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.727571939144685e-06,
+      "loss": 0.6774,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7227175382750145e-06,
+      "loss": 0.7803,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7178655773652283e-06,
+      "loss": 0.7266,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7130160583012666e-06,
+      "loss": 0.7474,
+      "step": 3766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708168982968118e-06,
+      "loss": 0.7436,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.703324353249833e-06,
+      "loss": 0.7733,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698482171029502e-06,
+      "loss": 0.8454,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.69364243818926e-06,
+      "loss": 0.7949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6888051566103044e-06,
+      "loss": 0.7675,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6839703281728677e-06,
+      "loss": 0.8113,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679137954756232e-06,
+      "loss": 0.793,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.674308038238725e-06,
+      "loss": 0.8207,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6694805804977183e-06,
+      "loss": 0.7244,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6646555834096355e-06,
+      "loss": 0.8058,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.659833048849929e-06,
+      "loss": 0.6704,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.655012978693113e-06,
+      "loss": 0.7054,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6501953748127283e-06,
+      "loss": 0.7307,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6453802390813586e-06,
+      "loss": 0.7675,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.640567573370639e-06,
+      "loss": 0.7316,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6357573795512347e-06,
+      "loss": 0.7691,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630949659492853e-06,
+      "loss": 0.7927,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6261444150642387e-06,
+      "loss": 0.781,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.621341648133172e-06,
+      "loss": 0.7858,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.61654136056648e-06,
+      "loss": 0.8118,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.611743554230013e-06,
+      "loss": 0.7162,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6069482309886695e-06,
+      "loss": 0.7748,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6021553927063725e-06,
+      "loss": 0.7471,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.597365041246085e-06,
+      "loss": 0.7647,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5925771784697993e-06,
+      "loss": 0.7755,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.587791806238541e-06,
+      "loss": 0.8269,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.583008926412375e-06,
+      "loss": 0.7466,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578228540850388e-06,
+      "loss": 0.3289,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5734506514106983e-06,
+      "loss": 0.7543,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5686752599504624e-06,
+      "loss": 0.7987,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563902368325853e-06,
+      "loss": 0.7748,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559131978392086e-06,
+      "loss": 0.8042,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5543640920033928e-06,
+      "loss": 0.725,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.549598711013035e-06,
+      "loss": 0.779,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5448358372733018e-06,
+      "loss": 0.7933,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540075472635505e-06,
+      "loss": 0.7901,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5353176189499873e-06,
+      "loss": 0.7832,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5305622780661074e-06,
+      "loss": 0.7749,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525809451832257e-06,
+      "loss": 0.7765,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5210591420958408e-06,
+      "loss": 0.785,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5163113507032898e-06,
+      "loss": 0.7879,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511566079500055e-06,
+      "loss": 0.7073,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5068233303306056e-06,
+      "loss": 0.7843,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5020831050384395e-06,
+      "loss": 0.3358,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.497345405466065e-06,
+      "loss": 0.6974,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.492610233455006e-06,
+      "loss": 0.747,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4878775908458174e-06,
+      "loss": 0.6472,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483147479478054e-06,
+      "loss": 0.7637,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478419901190305e-06,
+      "loss": 0.8259,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4736948578201534e-06,
+      "loss": 0.3439,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4689723512042183e-06,
+      "loss": 0.7992,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4642523831781206e-06,
+      "loss": 0.7379,
+      "step": 3818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4595349555764914e-06,
+      "loss": 0.7816,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.454820070232987e-06,
+      "loss": 0.7262,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4501077289802665e-06,
+      "loss": 0.8117,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4453979336499975e-06,
+      "loss": 0.7444,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4406906860728705e-06,
+      "loss": 0.7258,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.435985988078574e-06,
+      "loss": 0.7174,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4312838414958084e-06,
+      "loss": 0.7681,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4265842481522817e-06,
+      "loss": 0.3276,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4218872098747147e-06,
+      "loss": 0.812,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4171927284888307e-06,
+      "loss": 0.3539,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4125008058193564e-06,
+      "loss": 0.7215,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4078114436900323e-06,
+      "loss": 0.7895,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4031246439235965e-06,
+      "loss": 0.7769,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.398440408341792e-06,
+      "loss": 0.6925,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393758738765368e-06,
+      "loss": 0.8243,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389079637014071e-06,
+      "loss": 0.7749,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3844031049066594e-06,
+      "loss": 0.7452,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3797291442608805e-06,
+      "loss": 0.7368,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3750577568934937e-06,
+      "loss": 0.7496,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3703889446202497e-06,
+      "loss": 0.7573,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3657227092558975e-06,
+      "loss": 0.7271,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3610590526141938e-06,
+      "loss": 0.757,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356397976507887e-06,
+      "loss": 0.8242,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3517394827487194e-06,
+      "loss": 0.81,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347083573147434e-06,
+      "loss": 0.7533,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342430249513765e-06,
+      "loss": 0.7749,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3377795136564515e-06,
+      "loss": 0.7335,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3331313673832132e-06,
+      "loss": 0.7513,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328485812500776e-06,
+      "loss": 0.7994,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323842850814849e-06,
+      "loss": 0.7889,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3192024841301373e-06,
+      "loss": 0.7878,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.314564714250337e-06,
+      "loss": 0.3177,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3099295429781296e-06,
+      "loss": 0.7858,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3052969721152007e-06,
+      "loss": 0.7745,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300667003462208e-06,
+      "loss": 0.7458,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2960396388188122e-06,
+      "loss": 0.7695,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.291414879983652e-06,
+      "loss": 0.7781,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2867927287543535e-06,
+      "loss": 0.7527,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282173186927539e-06,
+      "loss": 0.7211,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.277556256298806e-06,
+      "loss": 0.7988,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2729419386627414e-06,
+      "loss": 0.7375,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2683302358129164e-06,
+      "loss": 0.8139,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263721149541881e-06,
+      "loss": 0.7989,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.25911468164118e-06,
+      "loss": 0.8181,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2545108339013254e-06,
+      "loss": 0.3416,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2499096081118254e-06,
+      "loss": 0.7557,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.245311006061159e-06,
+      "loss": 0.746,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240715029536787e-06,
+      "loss": 0.7326,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2361216803251528e-06,
+      "loss": 0.7806,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.231530960211672e-06,
+      "loss": 0.7448,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2269428709807505e-06,
+      "loss": 0.7974,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22235741441576e-06,
+      "loss": 0.7675,
+      "step": 3870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217774592299051e-06,
+      "loss": 0.8136,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213194406411957e-06,
+      "loss": 0.7845,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2086168585347756e-06,
+      "loss": 0.6847,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.204041950446792e-06,
+      "loss": 0.7687,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1994696839262562e-06,
+      "loss": 0.7536,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.194900060750393e-06,
+      "loss": 0.7396,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1903330826954006e-06,
+      "loss": 0.7158,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1857687515364445e-06,
+      "loss": 0.8125,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.181207069047674e-06,
+      "loss": 0.732,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1766480370021967e-06,
+      "loss": 0.8134,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1720916571720915e-06,
+      "loss": 0.7894,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.167537931328414e-06,
+      "loss": 0.7352,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162986861241181e-06,
+      "loss": 0.736,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1584384486793774e-06,
+      "loss": 0.7904,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.153892695410956e-06,
+      "loss": 0.68,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1493496032028416e-06,
+      "loss": 0.8406,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1448091738209165e-06,
+      "loss": 0.7682,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140271409030029e-06,
+      "loss": 0.787,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1357363105939997e-06,
+      "loss": 0.7886,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131203880275604e-06,
+      "loss": 0.7151,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1266741198365834e-06,
+      "loss": 0.693,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1221470310376423e-06,
+      "loss": 0.8372,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1176226156384405e-06,
+      "loss": 0.778,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1131008753976123e-06,
+      "loss": 0.7601,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1085818120727363e-06,
+      "loss": 0.8266,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104065427420364e-06,
+      "loss": 0.7615,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0995517231959994e-06,
+      "loss": 0.7762,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0950407011540996e-06,
+      "loss": 0.7775,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0905323630480912e-06,
+      "loss": 0.3517,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0860267106303487e-06,
+      "loss": 0.772,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081523745652204e-06,
+      "loss": 0.748,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0770234698639434e-06,
+      "loss": 0.7634,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.072525885014814e-06,
+      "loss": 0.7458,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068030992853013e-06,
+      "loss": 0.7738,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0635387951256846e-06,
+      "loss": 0.7348,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0590492935789396e-06,
+      "loss": 0.8092,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.054562489957831e-06,
+      "loss": 0.8469,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050078386006362e-06,
+      "loss": 0.7392,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0455969834674926e-06,
+      "loss": 0.7737,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0411182840831255e-06,
+      "loss": 0.7519,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.036642289594124e-06,
+      "loss": 0.8181,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032169001740285e-06,
+      "loss": 0.7803,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.02769842226037e-06,
+      "loss": 0.8002,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0232305528920748e-06,
+      "loss": 0.7508,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.018765395372042e-06,
+      "loss": 0.7815,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0143029514358725e-06,
+      "loss": 0.6945,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0098432228180994e-06,
+      "loss": 0.6993,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005386211252207e-06,
+      "loss": 0.7315,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.00093191847062e-06,
+      "loss": 0.8175,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9964803462047055e-06,
+      "loss": 0.7959,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.992031496184781e-06,
+      "loss": 0.8159,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9875853701400947e-06,
+      "loss": 0.7953,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9831419697988486e-06,
+      "loss": 0.7805,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9787012968881755e-06,
+      "loss": 0.7784,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9742633531341493e-06,
+      "loss": 0.7959,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9698281402617856e-06,
+      "loss": 0.786,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965395659995035e-06,
+      "loss": 0.7995,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9609659140567947e-06,
+      "loss": 0.7828,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9565389041688896e-06,
+      "loss": 0.7283,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.952114632052081e-06,
+      "loss": 0.7479,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9476930994260764e-06,
+      "loss": 0.7397,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.943274308009505e-06,
+      "loss": 0.7985,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9388582595199442e-06,
+      "loss": 0.8364,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934444955673893e-06,
+      "loss": 0.7772,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9300343981867906e-06,
+      "loss": 0.7378,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.925626588773005e-06,
+      "loss": 0.7019,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9212215291458368e-06,
+      "loss": 0.7147,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9168192210175238e-06,
+      "loss": 0.6943,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9124196660992255e-06,
+      "loss": 0.742,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9080228661010333e-06,
+      "loss": 0.7598,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903628822731975e-06,
+      "loss": 0.7697,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899237537699998e-06,
+      "loss": 0.7462,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894849012711982e-06,
+      "loss": 0.7155,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8904632494737283e-06,
+      "loss": 0.761,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8860802496899765e-06,
+      "loss": 0.7954,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8817000150643813e-06,
+      "loss": 0.761,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8773225472995224e-06,
+      "loss": 0.8057,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8729478480969154e-06,
+      "loss": 0.781,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8685759191569852e-06,
+      "loss": 0.7515,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8642067621790926e-06,
+      "loss": 0.7055,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.859840378861515e-06,
+      "loss": 0.7518,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855476770901442e-06,
+      "loss": 0.7026,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851115939995005e-06,
+      "loss": 0.7581,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8467578878372394e-06,
+      "loss": 0.8049,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.84240261612211e-06,
+      "loss": 0.7677,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8380501265424963e-06,
+      "loss": 0.7676,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.833700420790193e-06,
+      "loss": 0.8,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8293535005559237e-06,
+      "loss": 0.3439,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82500936752932e-06,
+      "loss": 0.7314,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8206680233989314e-06,
+      "loss": 0.7277,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8163294698522224e-06,
+      "loss": 0.6943,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811993708575581e-06,
+      "loss": 0.8167,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.807660741254302e-06,
+      "loss": 0.7563,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.803330569572591e-06,
+      "loss": 0.7889,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79900319521358e-06,
+      "loss": 0.7913,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.794678619859301e-06,
+      "loss": 0.8243,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7903568451907036e-06,
+      "loss": 0.726,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7860378728876493e-06,
+      "loss": 0.7326,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7817217046289037e-06,
+      "loss": 0.717,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777408342092155e-06,
+      "loss": 0.7439,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7730977869539878e-06,
+      "loss": 0.7645,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768790040889906e-06,
+      "loss": 0.7307,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764485105574316e-06,
+      "loss": 0.7602,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760182982680528e-06,
+      "loss": 0.7802,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7558836738807704e-06,
+      "loss": 0.6812,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.751587180846167e-06,
+      "loss": 0.7619,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.747293505246752e-06,
+      "loss": 0.6922,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7430026487514628e-06,
+      "loss": 0.7879,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7387146130281394e-06,
+      "loss": 0.7687,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.734429399743533e-06,
+      "loss": 0.7753,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7301470105632875e-06,
+      "loss": 0.7364,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725867447151961e-06,
+      "loss": 0.3441,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.721590711173e-06,
+      "loss": 0.8087,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.717316804288761e-06,
+      "loss": 0.7747,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713045728160496e-06,
+      "loss": 0.7857,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708777484448358e-06,
+      "loss": 0.8174,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.704512074811405e-06,
+      "loss": 0.8442,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7002495009075856e-06,
+      "loss": 0.7414,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6959897643937448e-06,
+      "loss": 0.723,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.691732866925636e-06,
+      "loss": 0.3213,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.687478810157895e-06,
+      "loss": 0.7676,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6832275957440658e-06,
+      "loss": 0.7457,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67897922533658e-06,
+      "loss": 0.7669,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674733700586766e-06,
+      "loss": 0.8354,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6704910231448446e-06,
+      "loss": 0.7533,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.666251194659929e-06,
+      "loss": 0.3186,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6620142167800335e-06,
+      "loss": 0.7272,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65778009115205e-06,
+      "loss": 0.7902,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6535488194217786e-06,
+      "loss": 0.7532,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649320403233897e-06,
+      "loss": 0.3369,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6450948442319767e-06,
+      "loss": 0.7113,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6408721440584793e-06,
+      "loss": 0.8066,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6366523043547534e-06,
+      "loss": 0.7959,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6324353267610413e-06,
+      "loss": 0.7987,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6282212129164676e-06,
+      "loss": 0.6537,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.624009964459041e-06,
+      "loss": 0.7854,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.619801583025666e-06,
+      "loss": 0.7118,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615596070252123e-06,
+      "loss": 0.3146,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6113934277730866e-06,
+      "loss": 0.7449,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607193657222108e-06,
+      "loss": 0.703,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602996760231624e-06,
+      "loss": 0.7639,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5988027384329562e-06,
+      "loss": 0.798,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594611593456303e-06,
+      "loss": 0.8198,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5904233269307566e-06,
+      "loss": 0.7651,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5862379404842795e-06,
+      "loss": 0.7717,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5820554357437166e-06,
+      "loss": 0.8008,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5778758143347983e-06,
+      "loss": 0.7777,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5736990778821268e-06,
+      "loss": 0.7549,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5695252280091888e-06,
+      "loss": 0.7782,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5653542663383426e-06,
+      "loss": 0.3202,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5611861944908334e-06,
+      "loss": 0.7305,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5570210140867747e-06,
+      "loss": 0.8153,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5528587267451554e-06,
+      "loss": 0.3198,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5486993340838505e-06,
+      "loss": 0.7026,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5445428377196e-06,
+      "loss": 0.7205,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5403892392680197e-06,
+      "loss": 0.7043,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5362385403436006e-06,
+      "loss": 0.681,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532090742559704e-06,
+      "loss": 0.7143,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5279458475285714e-06,
+      "loss": 0.7983,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5238038568613045e-06,
+      "loss": 0.3395,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5196647721678878e-06,
+      "loss": 0.8258,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5155285950571675e-06,
+      "loss": 0.7345,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5113953271368584e-06,
+      "loss": 0.6813,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5072649700135575e-06,
+      "loss": 0.765,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503137525292716e-06,
+      "loss": 0.7315,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49901299457866e-06,
+      "loss": 0.7421,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494891379474581e-06,
+      "loss": 0.7899,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.490772681582535e-06,
+      "loss": 0.7083,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4866569025034527e-06,
+      "loss": 0.747,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4825440438371186e-06,
+      "loss": 0.7472,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478434107182194e-06,
+      "loss": 0.8389,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4743270941361953e-06,
+      "loss": 0.7983,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4702230062955055e-06,
+      "loss": 0.804,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.46612184525537e-06,
+      "loss": 0.8268,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.462023612609895e-06,
+      "loss": 0.8327,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4579283099520556e-06,
+      "loss": 0.7012,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4538359388736775e-06,
+      "loss": 0.7639,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4497465009654597e-06,
+      "loss": 0.7828,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4456599978169495e-06,
+      "loss": 0.7728,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4415764310165546e-06,
+      "loss": 0.7116,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4374958021515505e-06,
+      "loss": 0.77,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4334181128080626e-06,
+      "loss": 0.8024,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429343364571075e-06,
+      "loss": 0.3155,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4252715590244303e-06,
+      "loss": 0.7313,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.421202697750822e-06,
+      "loss": 0.6957,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4171367823318115e-06,
+      "loss": 0.7612,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4130738143478005e-06,
+      "loss": 0.7292,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409013795378058e-06,
+      "loss": 0.7528,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404956727000698e-06,
+      "loss": 0.697,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400902610792688e-06,
+      "loss": 0.7897,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396851448329853e-06,
+      "loss": 0.7274,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392803241186862e-06,
+      "loss": 0.7731,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3887579909372472e-06,
+      "loss": 0.7655,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3847156991533825e-06,
+      "loss": 0.7734,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3806763674064893e-06,
+      "loss": 0.745,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3766399972666498e-06,
+      "loss": 0.7626,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3726065903027816e-06,
+      "loss": 0.7608,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3685761480826653e-06,
+      "loss": 0.733,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364548672172916e-06,
+      "loss": 0.7735,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360524164139002e-06,
+      "loss": 0.7533,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3565026255452363e-06,
+      "loss": 0.7802,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352484057954776e-06,
+      "loss": 0.3285,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3484684629296307e-06,
+      "loss": 0.7106,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3444558420306464e-06,
+      "loss": 0.7973,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340446196817515e-06,
+      "loss": 0.7927,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336439528848776e-06,
+      "loss": 0.7618,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332435839681808e-06,
+      "loss": 0.371,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3284351308728325e-06,
+      "loss": 0.7954,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324437403976908e-06,
+      "loss": 0.7237,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3204426605479448e-06,
+      "loss": 0.7656,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3164509021386828e-06,
+      "loss": 0.7186,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3124621303007055e-06,
+      "loss": 0.7713,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308476346584441e-06,
+      "loss": 0.739,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3044935525391454e-06,
+      "loss": 0.7924,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3005137497129193e-06,
+      "loss": 0.7402,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2965369396527048e-06,
+      "loss": 0.7311,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2925631239042666e-06,
+      "loss": 0.6996,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2885923040122214e-06,
+      "loss": 0.7595,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2846244815200082e-06,
+      "loss": 0.7151,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280659657969914e-06,
+      "loss": 0.7572,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.27669783490305e-06,
+      "loss": 0.761,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272739013859361e-06,
+      "loss": 0.7731,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268783196377634e-06,
+      "loss": 0.7574,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.26483038399548e-06,
+      "loss": 0.3028,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2608805782493448e-06,
+      "loss": 0.7852,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256933780674504e-06,
+      "loss": 0.782,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.252989992805068e-06,
+      "loss": 0.3652,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2490492161739753e-06,
+      "loss": 0.8108,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.245111452312987e-06,
+      "loss": 0.7416,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2411767027527076e-06,
+      "loss": 0.7178,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2372449690225585e-06,
+      "loss": 0.7127,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2333162526507923e-06,
+      "loss": 0.7768,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229390555164488e-06,
+      "loss": 0.773,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225467878089549e-06,
+      "loss": 0.7311,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2215482229507145e-06,
+      "loss": 0.768,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217631591271535e-06,
+      "loss": 0.3312,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2137179845743986e-06,
+      "loss": 0.731,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2098074043805094e-06,
+      "loss": 0.7603,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.205899852209895e-06,
+      "loss": 0.7576,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2019953295814146e-06,
+      "loss": 0.7599,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1980938380127394e-06,
+      "loss": 0.7576,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1941953790203686e-06,
+      "loss": 0.739,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19029995411962e-06,
+      "loss": 0.3112,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1864075648246296e-06,
+      "loss": 0.7615,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1825182126483646e-06,
+      "loss": 0.7716,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1786318991025956e-06,
+      "loss": 0.7903,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.174748625697928e-06,
+      "loss": 0.7414,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1708683939437745e-06,
+      "loss": 0.7121,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166991205348369e-06,
+      "loss": 0.7811,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1631170614187613e-06,
+      "loss": 0.7093,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1592459636608187e-06,
+      "loss": 0.6821,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1553779135792274e-06,
+      "loss": 0.7611,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1515129126774856e-06,
+      "loss": 0.7705,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1476509624579035e-06,
+      "loss": 0.3191,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1437920644216138e-06,
+      "loss": 0.7217,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1399362200685535e-06,
+      "loss": 0.7189,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1360834308974822e-06,
+      "loss": 0.78,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132233698405963e-06,
+      "loss": 0.786,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128387024090377e-06,
+      "loss": 0.3152,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1245434094459137e-06,
+      "loss": 0.7421,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.12070285596657e-06,
+      "loss": 0.7891,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1168653651451633e-06,
+      "loss": 0.7434,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1130309384733116e-06,
+      "loss": 0.7548,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109199577441441e-06,
+      "loss": 0.8135,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1053712835387963e-06,
+      "loss": 0.7328,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.101546058253419e-06,
+      "loss": 0.7641,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.097723903072162e-06,
+      "loss": 0.6771,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0939048194806833e-06,
+      "loss": 0.7333,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090088808963454e-06,
+      "loss": 0.7057,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0862758730037425e-06,
+      "loss": 0.7009,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0824660130836215e-06,
+      "loss": 0.7821,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.078659230683978e-06,
+      "loss": 0.753,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0748555272844904e-06,
+      "loss": 0.7563,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071054904363652e-06,
+      "loss": 0.7318,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0672573633987493e-06,
+      "loss": 0.773,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063462905865875e-06,
+      "loss": 0.8084,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.059671533239924e-06,
+      "loss": 0.752,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0558832469945853e-06,
+      "loss": 0.8082,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0520980486023623e-06,
+      "loss": 0.7662,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0483159395345454e-06,
+      "loss": 0.7359,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0445369212612255e-06,
+      "loss": 0.7839,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0407609952513008e-06,
+      "loss": 0.7753,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036988162972459e-06,
+      "loss": 0.7541,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0332184258911893e-06,
+      "loss": 0.7528,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0294517854727724e-06,
+      "loss": 0.7375,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0256882431812953e-06,
+      "loss": 0.8303,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021927800479633e-06,
+      "loss": 0.702,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.018170458829454e-06,
+      "loss": 0.7466,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0144162196912308e-06,
+      "loss": 0.7308,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0106650845242224e-06,
+      "loss": 0.7599,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0069170547864803e-06,
+      "loss": 0.7111,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0031721319348596e-06,
+      "loss": 0.7241,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994303174249887e-06,
+      "loss": 0.8199,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995691612711308e-06,
+      "loss": 0.787,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9919560192470346e-06,
+      "loss": 0.8134,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9882235384841864e-06,
+      "loss": 0.7154,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9844941718735653e-06,
+      "loss": 0.8026,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9807679208647625e-06,
+      "loss": 0.8173,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9770447869061625e-06,
+      "loss": 0.8325,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733247714449354e-06,
+      "loss": 0.7907,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9696078759270386e-06,
+      "loss": 0.7239,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9658941017972187e-06,
+      "loss": 0.7373,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9621834504990044e-06,
+      "loss": 0.7014,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9584759234747187e-06,
+      "loss": 0.7629,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9547715221654617e-06,
+      "loss": 0.7923,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951070248011128e-06,
+      "loss": 0.7863,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9473721024503865e-06,
+      "loss": 0.8106,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943677086920698e-06,
+      "loss": 0.7747,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9399852028583e-06,
+      "loss": 0.7345,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9362964516982143e-06,
+      "loss": 0.7387,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9326108348742545e-06,
+      "loss": 0.7834,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.928928353819003e-06,
+      "loss": 0.753,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9252490099638265e-06,
+      "loss": 0.7955,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9215728047388804e-06,
+      "loss": 0.7563,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917899739573088e-06,
+      "loss": 0.7762,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914229815894164e-06,
+      "loss": 0.7953,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910563035128593e-06,
+      "loss": 0.7689,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9068993987016415e-06,
+      "loss": 0.7735,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9032389080373537e-06,
+      "loss": 0.3177,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.899581564558548e-06,
+      "loss": 0.7433,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8959273696868264e-06,
+      "loss": 0.6804,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.892276324842558e-06,
+      "loss": 0.8546,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8886284314449e-06,
+      "loss": 0.8229,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8849836909117713e-06,
+      "loss": 0.7753,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8813421046598723e-06,
+      "loss": 0.8237,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8777036741046761e-06,
+      "loss": 0.7583,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874068400660427e-06,
+      "loss": 0.7399,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8704362857401492e-06,
+      "loss": 0.7692,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.866807330755631e-06,
+      "loss": 0.7653,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8631815371174344e-06,
+      "loss": 0.7569,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8595589062348985e-06,
+      "loss": 0.7727,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8559394395161223e-06,
+      "loss": 0.3228,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8523231383679885e-06,
+      "loss": 0.7868,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8487100041961382e-06,
+      "loss": 0.7297,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8451000384049845e-06,
+      "loss": 0.8103,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8414932423977105e-06,
+      "loss": 0.7343,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837889617576264e-06,
+      "loss": 0.3075,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8342891653413676e-06,
+      "loss": 0.3138,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8306918870925028e-06,
+      "loss": 0.7614,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827097784227917e-06,
+      "loss": 0.7229,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8235068581446348e-06,
+      "loss": 0.8131,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8199191102384317e-06,
+      "loss": 0.7666,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8163345419038559e-06,
+      "loss": 0.7964,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8127531545342147e-06,
+      "loss": 0.7115,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809174949521587e-06,
+      "loss": 0.7651,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8055999282568092e-06,
+      "loss": 0.7489,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8020280921294752e-06,
+      "loss": 0.6921,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7984594425279534e-06,
+      "loss": 0.7729,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7948939808393628e-06,
+      "loss": 0.7943,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7913317084495862e-06,
+      "loss": 0.7131,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7877726267432749e-06,
+      "loss": 0.7553,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784216737103821e-06,
+      "loss": 0.731,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7806640409133969e-06,
+      "loss": 0.7976,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7771145395529188e-06,
+      "loss": 0.7816,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7735682344020722e-06,
+      "loss": 0.8324,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.770025126839291e-06,
+      "loss": 0.762,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.766485218241768e-06,
+      "loss": 0.8333,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7629485099854604e-06,
+      "loss": 0.7996,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7594150034450709e-06,
+      "loss": 0.7461,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7558846999940648e-06,
+      "loss": 0.7094,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7523576010046585e-06,
+      "loss": 0.3219,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7488337078478212e-06,
+      "loss": 0.7746,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7453130218932858e-06,
+      "loss": 0.7118,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7417955445095258e-06,
+      "loss": 0.7378,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7382812770637791e-06,
+      "loss": 0.2938,
+      "step": 4235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7347702209220264e-06,
+      "loss": 0.71,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7312623774490057e-06,
+      "loss": 0.7959,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7277577480082042e-06,
+      "loss": 0.7385,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7242563339618578e-06,
+      "loss": 0.7763,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7207581366709603e-06,
+      "loss": 0.7648,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7172631574952448e-06,
+      "loss": 0.7035,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7137713977932024e-06,
+      "loss": 0.7797,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7102828589220689e-06,
+      "loss": 0.8147,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067975422378247e-06,
+      "loss": 0.6634,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7033154490952064e-06,
+      "loss": 0.3459,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6998365808476891e-06,
+      "loss": 0.3144,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6963609388475e-06,
+      "loss": 0.7355,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6928885244456084e-06,
+      "loss": 0.7877,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.689419338991729e-06,
+      "loss": 0.7755,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.685953383834329e-06,
+      "loss": 0.6953,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.682490660320608e-06,
+      "loss": 0.78,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6790311697965212e-06,
+      "loss": 0.7709,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.675574913606759e-06,
+      "loss": 0.7428,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672121893094758e-06,
+      "loss": 0.7391,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6686721096026959e-06,
+      "loss": 0.7212,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6652255644714888e-06,
+      "loss": 0.6952,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661782259040805e-06,
+      "loss": 0.8153,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.658342194649043e-06,
+      "loss": 0.7444,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.654905372633343e-06,
+      "loss": 0.7983,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.651471794329591e-06,
+      "loss": 0.7846,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6480414610724038e-06,
+      "loss": 0.7228,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.644614374195146e-06,
+      "loss": 0.8283,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6411905350299118e-06,
+      "loss": 0.7633,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.637769944907539e-06,
+      "loss": 0.7733,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6343526051575976e-06,
+      "loss": 0.7152,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.630938517108397e-06,
+      "loss": 0.2995,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6275276820869845e-06,
+      "loss": 0.8348,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6241201014191399e-06,
+      "loss": 0.7558,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.620715776429378e-06,
+      "loss": 0.8313,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6173147084409513e-06,
+      "loss": 0.3382,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6139168987758437e-06,
+      "loss": 0.7365,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6105223487547727e-06,
+      "loss": 0.7526,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6071310596971856e-06,
+      "loss": 0.7764,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6037430329212722e-06,
+      "loss": 0.789,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6003582697439447e-06,
+      "loss": 0.721,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5969767714808481e-06,
+      "loss": 0.7772,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5935985394463637e-06,
+      "loss": 0.7878,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5902235749535977e-06,
+      "loss": 0.7453,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.586851879314386e-06,
+      "loss": 0.7975,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5834834538393029e-06,
+      "loss": 0.7778,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5801182998376363e-06,
+      "loss": 0.7176,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5767564186174167e-06,
+      "loss": 0.7737,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.573397811485392e-06,
+      "loss": 0.7431,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5700424797470482e-06,
+      "loss": 0.8105,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5666904247065885e-06,
+      "loss": 0.8076,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.563341647666945e-06,
+      "loss": 0.8133,
+      "step": 4286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.55999614992978e-06,
+      "loss": 0.7733,
+      "step": 4287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5566539327954766e-06,
+      "loss": 0.7608,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5533149975631445e-06,
+      "loss": 0.7448,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5499793455306133e-06,
+      "loss": 0.8475,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5466469779944459e-06,
+      "loss": 0.7442,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5433178962499207e-06,
+      "loss": 0.7297,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5399921015910402e-06,
+      "loss": 0.7962,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5366695953105326e-06,
+      "loss": 0.7536,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5333503786998449e-06,
+      "loss": 0.7453,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5300344530491428e-06,
+      "loss": 0.7439,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5267218196473254e-06,
+      "loss": 0.726,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5234124797819905e-06,
+      "loss": 0.7648,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5201064347394768e-06,
+      "loss": 0.7375,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5168036858048274e-06,
+      "loss": 0.7694,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5135042342618178e-06,
+      "loss": 0.7372,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5102080813929298e-06,
+      "loss": 0.8062,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5069152284793653e-06,
+      "loss": 0.7385,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5036256768010516e-06,
+      "loss": 0.7667,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5003394276366245e-06,
+      "loss": 0.7323,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4970564822634382e-06,
+      "loss": 0.7154,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4937768419575626e-06,
+      "loss": 0.7429,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4905005079937829e-06,
+      "loss": 0.8053,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4872274816456034e-06,
+      "loss": 0.746,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4839577641852342e-06,
+      "loss": 0.7033,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.48069135688361e-06,
+      "loss": 0.7997,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4774282610103695e-06,
+      "loss": 0.7917,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4741684778338672e-06,
+      "loss": 0.8023,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4709120086211726e-06,
+      "loss": 0.6983,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.467658854638062e-06,
+      "loss": 0.2992,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4644090171490298e-06,
+      "loss": 0.6961,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4611624974172756e-06,
+      "loss": 0.7307,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4579192967047085e-06,
+      "loss": 0.7604,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4546794162719568e-06,
+      "loss": 0.7666,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4514428573783456e-06,
+      "loss": 0.3434,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4482096212819208e-06,
+      "loss": 0.8059,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444979709239428e-06,
+      "loss": 0.6904,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4417531225063242e-06,
+      "loss": 0.7152,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4385298623367727e-06,
+      "loss": 0.7662,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4353099299836437e-06,
+      "loss": 0.7548,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4320933266985182e-06,
+      "loss": 0.7754,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4288800537316772e-06,
+      "loss": 0.7461,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4256701123321082e-06,
+      "loss": 0.6922,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.42246350374751e-06,
+      "loss": 0.7179,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4192602292242786e-06,
+      "loss": 0.745,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416060290007517e-06,
+      "loss": 0.803,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4128636873410283e-06,
+      "loss": 0.7281,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4096704224673285e-06,
+      "loss": 0.8072,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4064804966276247e-06,
+      "loss": 0.8203,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4032939110618294e-06,
+      "loss": 0.7784,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400110667008564e-06,
+      "loss": 0.7499,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3969307657051412e-06,
+      "loss": 0.7202,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3937542083875811e-06,
+      "loss": 0.674,
+      "step": 4338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3905809962906002e-06,
+      "loss": 0.7724,
+      "step": 4339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3874111306476168e-06,
+      "loss": 0.7387,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3842446126907472e-06,
+      "loss": 0.7226,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3810814436508046e-06,
+      "loss": 0.8194,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3779216247573068e-06,
+      "loss": 0.7319,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3747651572384636e-06,
+      "loss": 0.739,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.371612042321182e-06,
+      "loss": 0.692,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3684622812310723e-06,
+      "loss": 0.7396,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3653158751924334e-06,
+      "loss": 0.7884,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3621728254282652e-06,
+      "loss": 0.7411,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3590331331602568e-06,
+      "loss": 0.8358,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3558967996088034e-06,
+      "loss": 0.3329,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3527638259929843e-06,
+      "loss": 0.7327,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3496342135305752e-06,
+      "loss": 0.758,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3465079634380496e-06,
+      "loss": 0.8047,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.34338507693057e-06,
+      "loss": 0.7319,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3402655552219902e-06,
+      "loss": 0.3383,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3371493995248642e-06,
+      "loss": 0.762,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3340366110504243e-06,
+      "loss": 0.787,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3309271910086074e-06,
+      "loss": 0.7898,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.327821140608031e-06,
+      "loss": 0.7743,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3247184610560115e-06,
+      "loss": 0.7765,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3216191535585487e-06,
+      "loss": 0.6868,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3185232193203311e-06,
+      "loss": 0.3132,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3154306595447441e-06,
+      "loss": 0.762,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3123414754338515e-06,
+      "loss": 0.7588,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3092556681884106e-06,
+      "loss": 0.7621,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3061732390078652e-06,
+      "loss": 0.7013,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3030941890903437e-06,
+      "loss": 0.7512,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3000185196326665e-06,
+      "loss": 0.7659,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2969462318303338e-06,
+      "loss": 0.2959,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2938773268775373e-06,
+      "loss": 0.7768,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.290811805967148e-06,
+      "loss": 0.7339,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2877496702907232e-06,
+      "loss": 0.7573,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2846909210385116e-06,
+      "loss": 0.326,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2816355593994312e-06,
+      "loss": 0.7837,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2785835865610974e-06,
+      "loss": 0.8323,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2755350037097991e-06,
+      "loss": 0.8086,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2724898120305107e-06,
+      "loss": 0.7063,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.269448012706892e-06,
+      "loss": 0.7599,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2664096069212762e-06,
+      "loss": 0.7503,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2633745958546862e-06,
+      "loss": 0.7921,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2603429806868206e-06,
+      "loss": 0.7103,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2573147625960569e-06,
+      "loss": 0.7379,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2542899427594545e-06,
+      "loss": 0.7599,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2512685223527487e-06,
+      "loss": 0.7458,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2482505025503611e-06,
+      "loss": 0.7362,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.245235884525383e-06,
+      "loss": 0.754,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2422246694495898e-06,
+      "loss": 0.705,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2392168584934283e-06,
+      "loss": 0.7373,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2362124528260278e-06,
+      "loss": 0.7174,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2332114536151895e-06,
+      "loss": 0.7676,
+      "step": 4390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2302138620273907e-06,
+      "loss": 0.8055,
+      "step": 4391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.227219679227789e-06,
+      "loss": 0.8088,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2242289063802137e-06,
+      "loss": 0.718,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.221241544647165e-06,
+      "loss": 0.781,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2182575951898257e-06,
+      "loss": 0.6368,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.215277059168044e-06,
+      "loss": 0.7597,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2122999377403478e-06,
+      "loss": 0.7293,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2093262320639333e-06,
+      "loss": 0.6454,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2063559432946715e-06,
+      "loss": 0.7397,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2033890725871022e-06,
+      "loss": 0.7099,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2004256210944377e-06,
+      "loss": 0.7811,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1974655899685671e-06,
+      "loss": 0.8384,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1945089803600419e-06,
+      "loss": 0.81,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1915557934180855e-06,
+      "loss": 0.8523,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1886060302905956e-06,
+      "loss": 0.7701,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1856596921241359e-06,
+      "loss": 0.7352,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1827167800639362e-06,
+      "loss": 0.8194,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1797772952538967e-06,
+      "loss": 0.7786,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176841238836589e-06,
+      "loss": 0.7611,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1739086119532472e-06,
+      "loss": 0.7287,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1709794157437738e-06,
+      "loss": 0.6903,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1680536513467389e-06,
+      "loss": 0.7503,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1651313198993786e-06,
+      "loss": 0.8147,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1622124225375909e-06,
+      "loss": 0.768,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1592969603959469e-06,
+      "loss": 0.7487,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1563849346076716e-06,
+      "loss": 0.6922,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1534763463046661e-06,
+      "loss": 0.3297,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1505711966174838e-06,
+      "loss": 0.8209,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1476694866753513e-06,
+      "loss": 0.3108,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1447712176061543e-06,
+      "loss": 0.8238,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141876390536436e-06,
+      "loss": 0.6701,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1389850065914133e-06,
+      "loss": 0.7409,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1360970668949534e-06,
+      "loss": 0.7359,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1332125725695918e-06,
+      "loss": 0.7363,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130331524736521e-06,
+      "loss": 0.7886,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1274539245155925e-06,
+      "loss": 0.7464,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124579773025326e-06,
+      "loss": 0.7562,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1217090713828904e-06,
+      "loss": 0.7725,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1188418207041218e-06,
+      "loss": 0.7205,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1159780221035111e-06,
+      "loss": 0.7481,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1131176766942042e-06,
+      "loss": 0.3333,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1102607855880155e-06,
+      "loss": 0.8116,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1074073498953996e-06,
+      "loss": 0.7396,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1045573707254864e-06,
+      "loss": 0.7501,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1017108491860485e-06,
+      "loss": 0.8238,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.098867786383524e-06,
+      "loss": 0.7852,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0960281834230002e-06,
+      "loss": 0.777,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0931920414082197e-06,
+      "loss": 0.7686,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0903593614415875e-06,
+      "loss": 0.7963,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0875301446241536e-06,
+      "loss": 0.7578,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0847043920556266e-06,
+      "loss": 0.8282,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0818821048343665e-06,
+      "loss": 0.7363,
+      "step": 4442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0790632840573867e-06,
+      "loss": 0.8156,
+      "step": 4443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0762479308203567e-06,
+      "loss": 0.8117,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073436046217593e-06,
+      "loss": 0.7751,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0706276313420693e-06,
+      "loss": 0.8419,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0678226872854058e-06,
+      "loss": 0.7236,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0650212151378736e-06,
+      "loss": 0.7305,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0622232159884004e-06,
+      "loss": 0.8556,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0594286909245543e-06,
+      "loss": 0.788,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.056637641032563e-06,
+      "loss": 0.7998,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0538500673972974e-06,
+      "loss": 0.7915,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0510659711022753e-06,
+      "loss": 0.7524,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.048285353229671e-06,
+      "loss": 0.72,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045508214860298e-06,
+      "loss": 0.7746,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0427345570736246e-06,
+      "loss": 0.7622,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0399643809477623e-06,
+      "loss": 0.7358,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.037197687559467e-06,
+      "loss": 0.6927,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0344344779841475e-06,
+      "loss": 0.7716,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0316747532958493e-06,
+      "loss": 0.6657,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0289185145672743e-06,
+      "loss": 0.7716,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026165762869762e-06,
+      "loss": 0.7507,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0234164992732954e-06,
+      "loss": 0.7206,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0206707248465086e-06,
+      "loss": 0.757,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0179284406566736e-06,
+      "loss": 0.7367,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0151896477697086e-06,
+      "loss": 0.757,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0124543472501713e-06,
+      "loss": 0.7668,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0097225401612675e-06,
+      "loss": 0.7558,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0069942275648403e-06,
+      "loss": 0.7518,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0042694105213747e-06,
+      "loss": 0.7937,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0015480900900021e-06,
+      "loss": 0.8066,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.988302673284877e-07,
+      "loss": 0.6958,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.961159432932443e-07,
+      "loss": 0.8131,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.9340511903932e-07,
+      "loss": 0.7037,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.906977956204e-07,
+      "loss": 0.7608,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.879939740888155e-07,
+      "loss": 0.3235,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.852936554955328e-07,
+      "loss": 0.7754,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.825968408901576e-07,
+      "loss": 0.3112,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.799035313209337e-07,
+      "loss": 0.7706,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.772137278347394e-07,
+      "loss": 0.7444,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.745274314770968e-07,
+      "loss": 0.7275,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.71844643292159e-07,
+      "loss": 0.7978,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69165364322716e-07,
+      "loss": 0.7805,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66489595610195e-07,
+      "loss": 0.7529,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.638173381946626e-07,
+      "loss": 0.7345,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.611485931148123e-07,
+      "loss": 0.7532,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58483361407978e-07,
+      "loss": 0.7398,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558216441101297e-07,
+      "loss": 0.7545,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.531634422558656e-07,
+      "loss": 0.7913,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.505087568784199e-07,
+      "loss": 0.7476,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.478575890096664e-07,
+      "loss": 0.7133,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.452099396800962e-07,
+      "loss": 0.7471,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.425658099188484e-07,
+      "loss": 0.7626,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.399252007536863e-07,
+      "loss": 0.8291,
+      "step": 4494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.372881132110068e-07,
+      "loss": 0.7561,
+      "step": 4495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.346545483158387e-07,
+      "loss": 0.7451,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.320245070918366e-07,
+      "loss": 0.8139,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.293979905612938e-07,
+      "loss": 0.7224,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.267749997451259e-07,
+      "loss": 0.7485,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.241555356628828e-07,
+      "loss": 0.6785,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.215395993327414e-07,
+      "loss": 0.74,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.189271917715048e-07,
+      "loss": 0.7452,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.16318313994613e-07,
+      "loss": 0.7015,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.137129670161238e-07,
+      "loss": 0.7096,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.111111518487314e-07,
+      "loss": 0.732,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.085128695037515e-07,
+      "loss": 0.7278,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.059181209911272e-07,
+      "loss": 0.7852,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.033269073194318e-07,
+      "loss": 0.7407,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.007392294958572e-07,
+      "loss": 0.7734,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.981550885262302e-07,
+      "loss": 0.8137,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.955744854149961e-07,
+      "loss": 0.761,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.92997421165227e-07,
+      "loss": 0.7492,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.904238967786205e-07,
+      "loss": 0.7566,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.878539132554953e-07,
+      "loss": 0.6756,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.852874715948001e-07,
+      "loss": 0.7262,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827245727941003e-07,
+      "loss": 0.7817,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.801652178495857e-07,
+      "loss": 0.6906,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.776094077560704e-07,
+      "loss": 0.7079,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750571435069888e-07,
+      "loss": 0.7563,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.725084260943995e-07,
+      "loss": 0.8162,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.699632565089811e-07,
+      "loss": 0.7524,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.674216357400289e-07,
+      "loss": 0.7619,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.648835647754694e-07,
+      "loss": 0.7349,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623490446018379e-07,
+      "loss": 0.7463,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.598180762042974e-07,
+      "loss": 0.7426,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.572906605666231e-07,
+      "loss": 0.7605,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.547667986712182e-07,
+      "loss": 0.7801,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.522464914990969e-07,
+      "loss": 0.7767,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.497297400298954e-07,
+      "loss": 0.7374,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.472165452418679e-07,
+      "loss": 0.3442,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.447069081118842e-07,
+      "loss": 0.7575,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.422008296154328e-07,
+      "loss": 0.7493,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.396983107266199e-07,
+      "loss": 0.3279,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.371993524181643e-07,
+      "loss": 0.7052,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.347039556614045e-07,
+      "loss": 0.7509,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.322121214262924e-07,
+      "loss": 0.7889,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.297238506813965e-07,
+      "loss": 0.7791,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.272391443939021e-07,
+      "loss": 0.784,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.24758003529601e-07,
+      "loss": 0.7399,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.222804290529108e-07,
+      "loss": 0.7067,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198064219268553e-07,
+      "loss": 0.7226,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.173359831130711e-07,
+      "loss": 0.8394,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.148691135718112e-07,
+      "loss": 0.7458,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124058142619407e-07,
+      "loss": 0.7125,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.099460861409359e-07,
+      "loss": 0.7941,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.074899301648842e-07,
+      "loss": 0.7638,
+      "step": 4546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.050373472884876e-07,
+      "loss": 0.7475,
+      "step": 4547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.025883384650568e-07,
+      "loss": 0.776,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.001429046465104e-07,
+      "loss": 0.7493,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.97701046783389e-07,
+      "loss": 0.7637,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.952627658248257e-07,
+      "loss": 0.7356,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.928280627185781e-07,
+      "loss": 0.7904,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.903969384110044e-07,
+      "loss": 0.7423,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.87969393847079e-07,
+      "loss": 0.7879,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.85545429970379e-07,
+      "loss": 0.7298,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.831250477230889e-07,
+      "loss": 0.773,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.807082480460082e-07,
+      "loss": 0.3599,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.782950318785365e-07,
+      "loss": 0.7565,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.758854001586835e-07,
+      "loss": 0.7035,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.734793538230656e-07,
+      "loss": 0.7565,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.710768938069046e-07,
+      "loss": 0.8065,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.686780210440315e-07,
+      "loss": 0.7803,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.662827364668768e-07,
+      "loss": 0.7879,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.638910410064848e-07,
+      "loss": 0.6991,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.615029355924974e-07,
+      "loss": 0.7768,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.591184211531622e-07,
+      "loss": 0.7389,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.567374986153375e-07,
+      "loss": 0.34,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.543601689044733e-07,
+      "loss": 0.7938,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.519864329446358e-07,
+      "loss": 0.3541,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.49616291658486e-07,
+      "loss": 0.7747,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.472497459672901e-07,
+      "loss": 0.7443,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.44886796790919e-07,
+      "loss": 0.7313,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.425274450478415e-07,
+      "loss": 0.8411,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.401716916551316e-07,
+      "loss": 0.7961,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.378195375284636e-07,
+      "loss": 0.8132,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.354709835821116e-07,
+      "loss": 0.7458,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.331260307289501e-07,
+      "loss": 0.3315,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.307846798804552e-07,
+      "loss": 0.3077,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284469319467036e-07,
+      "loss": 0.7411,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.261127878363694e-07,
+      "loss": 0.749,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.237822484567292e-07,
+      "loss": 0.8523,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.214553147136549e-07,
+      "loss": 0.7518,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.191319875116165e-07,
+      "loss": 0.7122,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.168122677536893e-07,
+      "loss": 0.7098,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.144961563415342e-07,
+      "loss": 0.7478,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.121836541754212e-07,
+      "loss": 0.7673,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.098747621542102e-07,
+      "loss": 0.7514,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.075694811753609e-07,
+      "loss": 0.741,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.052678121349299e-07,
+      "loss": 0.7645,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.029697559275662e-07,
+      "loss": 0.7242,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.006753134465194e-07,
+      "loss": 0.7971,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.98384485583632e-07,
+      "loss": 0.3259,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.960972732293414e-07,
+      "loss": 0.7261,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.938136772726778e-07,
+      "loss": 0.7501,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.915336986012677e-07,
+      "loss": 0.7646,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.892573381013357e-07,
+      "loss": 0.7508,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.869845966576927e-07,
+      "loss": 0.7824,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.847154751537443e-07,
+      "loss": 0.7662,
+      "step": 4598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.824499744714952e-07,
+      "loss": 0.7811,
+      "step": 4599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.801880954915374e-07,
+      "loss": 0.7372,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.779298390930544e-07,
+      "loss": 0.7297,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.75675206153823e-07,
+      "loss": 0.7328,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.734241975502143e-07,
+      "loss": 0.7831,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.711768141571861e-07,
+      "loss": 0.8159,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.689330568482888e-07,
+      "loss": 0.7482,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.666929264956679e-07,
+      "loss": 0.7224,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.644564239700513e-07,
+      "loss": 0.6925,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.622235501407593e-07,
+      "loss": 0.8027,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.599943058757085e-07,
+      "loss": 0.7397,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.577686920413928e-07,
+      "loss": 0.7498,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.555467095029045e-07,
+      "loss": 0.7791,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.533283591239203e-07,
+      "loss": 0.7673,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.511136417667075e-07,
+      "loss": 0.7063,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.489025582921183e-07,
+      "loss": 0.7204,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.46695109559593e-07,
+      "loss": 0.7454,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.444912964271633e-07,
+      "loss": 0.3426,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.422911197514414e-07,
+      "loss": 0.7693,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.400945803876302e-07,
+      "loss": 0.8004,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.379016791895176e-07,
+      "loss": 0.7273,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.357124170094753e-07,
+      "loss": 0.7341,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.335267946984658e-07,
+      "loss": 0.3194,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.313448131060307e-07,
+      "loss": 0.7209,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.291664730803026e-07,
+      "loss": 0.7583,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.269917754679944e-07,
+      "loss": 0.7875,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.248207211144008e-07,
+      "loss": 0.7069,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.226533108634102e-07,
+      "loss": 0.7659,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204895455574811e-07,
+      "loss": 0.752,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.183294260376671e-07,
+      "loss": 0.7699,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.161729531435967e-07,
+      "loss": 0.7541,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.140201277134872e-07,
+      "loss": 0.8211,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.118709505841324e-07,
+      "loss": 0.7928,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097254225909122e-07,
+      "loss": 0.8256,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.075835445677869e-07,
+      "loss": 0.7694,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.054453173472962e-07,
+      "loss": 0.6941,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.033107417605633e-07,
+      "loss": 0.3155,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.011798186372919e-07,
+      "loss": 0.7695,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.990525488057608e-07,
+      "loss": 0.7845,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.969289330928385e-07,
+      "loss": 0.7301,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.948089723239636e-07,
+      "loss": 0.7641,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.926926673231626e-07,
+      "loss": 0.7861,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.905800189130329e-07,
+      "loss": 0.8226,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.884710279147543e-07,
+      "loss": 0.7975,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.863656951480878e-07,
+      "loss": 0.7786,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.842640214313667e-07,
+      "loss": 0.7339,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821660075815061e-07,
+      "loss": 0.764,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.800716544139972e-07,
+      "loss": 0.3268,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.779809627429078e-07,
+      "loss": 0.7366,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.758939333808855e-07,
+      "loss": 0.795,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.73810567139148e-07,
+      "loss": 0.7586,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.717308648274977e-07,
+      "loss": 0.7017,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.696548272543068e-07,
+      "loss": 0.7366,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.67582455226523e-07,
+      "loss": 0.7326,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.655137495496721e-07,
+      "loss": 0.781,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.634487110278519e-07,
+      "loss": 0.7798,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.613873404637382e-07,
+      "loss": 0.7348,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5932963865858e-07,
+      "loss": 0.717,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.572756064121953e-07,
+      "loss": 0.7248,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.552252445229855e-07,
+      "loss": 0.701,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.531785537879141e-07,
+      "loss": 0.7509,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.511355350025305e-07,
+      "loss": 0.7862,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.490961889609425e-07,
+      "loss": 0.874,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.470605164558407e-07,
+      "loss": 0.7095,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.450285182784853e-07,
+      "loss": 0.7015,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.430001952187048e-07,
+      "loss": 0.724,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.409755480649059e-07,
+      "loss": 0.7618,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389545776040583e-07,
+      "loss": 0.7338,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.369372846217103e-07,
+      "loss": 0.7156,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.349236699019766e-07,
+      "loss": 0.7521,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.329137342275403e-07,
+      "loss": 0.7355,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.309074783796608e-07,
+      "loss": 0.7228,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.289049031381599e-07,
+      "loss": 0.8143,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.269060092814348e-07,
+      "loss": 0.7808,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.249107975864487e-07,
+      "loss": 0.8296,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.229192688287332e-07,
+      "loss": 0.7273,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.209314237823904e-07,
+      "loss": 0.8231,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.189472632200898e-07,
+      "loss": 0.8009,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.169667879130657e-07,
+      "loss": 0.7168,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.149899986311246e-07,
+      "loss": 0.746,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.130168961426385e-07,
+      "loss": 0.7048,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.110474812145461e-07,
+      "loss": 0.751,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.090817546123505e-07,
+      "loss": 0.8256,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.071197171001263e-07,
+      "loss": 0.7654,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.051613694405111e-07,
+      "loss": 0.6948,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.032067123947049e-07,
+      "loss": 0.8014,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.012557467224833e-07,
+      "loss": 0.8095,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.993084731821729e-07,
+      "loss": 0.7926,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.973648925306773e-07,
+      "loss": 0.7758,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.954250055234589e-07,
+      "loss": 0.6715,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.934888129145476e-07,
+      "loss": 0.7972,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.915563154565328e-07,
+      "loss": 0.7462,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896275139005702e-07,
+      "loss": 0.7774,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.877024089963822e-07,
+      "loss": 0.8008,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.85781001492247e-07,
+      "loss": 0.7697,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.838632921350128e-07,
+      "loss": 0.7159,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.819492816700844e-07,
+      "loss": 0.7422,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800389708414321e-07,
+      "loss": 0.7534,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7813236039159e-07,
+      "loss": 0.7287,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.762294510616483e-07,
+      "loss": 0.7302,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7433024359126443e-07,
+      "loss": 0.6916,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7243473871865455e-07,
+      "loss": 0.7588,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.705429371805925e-07,
+      "loss": 0.7869,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6865483971242023e-07,
+      "loss": 0.6395,
+      "step": 4702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6677044704802944e-07,
+      "loss": 0.7969,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6488975991988207e-07,
+      "loss": 0.7392,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6301277905899445e-07,
+      "loss": 0.7452,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.611395051949408e-07,
+      "loss": 0.7873,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.592699390558597e-07,
+      "loss": 0.7369,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.574040813684433e-07,
+      "loss": 0.811,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555419328579469e-07,
+      "loss": 0.7484,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.536834942481805e-07,
+      "loss": 0.7281,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5182876626151286e-07,
+      "loss": 0.6703,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4997774961887176e-07,
+      "loss": 0.7532,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4813044503973945e-07,
+      "loss": 0.7261,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.462868532421616e-07,
+      "loss": 0.8293,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.444469749427316e-07,
+      "loss": 0.7822,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.426108108566085e-07,
+      "loss": 0.8295,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4077836169750145e-07,
+      "loss": 0.815,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3894962817767614e-07,
+      "loss": 0.325,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3712461100796077e-07,
+      "loss": 0.7813,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.353033108977278e-07,
+      "loss": 0.6687,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.334857285549143e-07,
+      "loss": 0.7775,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.316718646860074e-07,
+      "loss": 0.7735,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2986171999605087e-07,
+      "loss": 0.739,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.280552951886419e-07,
+      "loss": 0.8164,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.26252590965931e-07,
+      "loss": 0.7935,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2445360802862657e-07,
+      "loss": 0.7568,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.22658347075986e-07,
+      "loss": 0.7168,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.208668088058221e-07,
+      "loss": 0.7721,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.190789939144979e-07,
+      "loss": 0.7805,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.172949030969331e-07,
+      "loss": 0.7301,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.155145370465985e-07,
+      "loss": 0.7147,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.137378964555172e-07,
+      "loss": 0.8109,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.119649820142613e-07,
+      "loss": 0.7245,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1019579441195945e-07,
+      "loss": 0.6863,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0843033433628945e-07,
+      "loss": 0.776,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0666860247347785e-07,
+      "loss": 0.6956,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.049105995083058e-07,
+      "loss": 0.7689,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.031563261241034e-07,
+      "loss": 0.7634,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.014057830027518e-07,
+      "loss": 0.7627,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.99658970824679e-07,
+      "loss": 0.804,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9791589026886845e-07,
+      "loss": 0.7667,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961765420128494e-07,
+      "loss": 0.7209,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9444092673269875e-07,
+      "loss": 0.7349,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.927090451030491e-07,
+      "loss": 0.7694,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909808977970719e-07,
+      "loss": 0.6981,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8925648548649664e-07,
+      "loss": 0.8111,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.875358088415937e-07,
+      "loss": 0.7579,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.858188685311892e-07,
+      "loss": 0.7199,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8410566522264824e-07,
+      "loss": 0.697,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.823961995818892e-07,
+      "loss": 0.754,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.806904722733762e-07,
+      "loss": 0.7749,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.78988483960121e-07,
+      "loss": 0.7393,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.772902353036789e-07,
+      "loss": 0.742,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7559572696415416e-07,
+      "loss": 0.7983,
+      "step": 4754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7390495960019645e-07,
+      "loss": 0.7756,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.722179338690046e-07,
+      "loss": 0.7758,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.705346504263163e-07,
+      "loss": 0.7435,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.688551099264226e-07,
+      "loss": 0.7013,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.671793130221524e-07,
+      "loss": 0.746,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6550726036488237e-07,
+      "loss": 0.8205,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.638389526045383e-07,
+      "loss": 0.6624,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.621743903895814e-07,
+      "loss": 0.7639,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6051357436702407e-07,
+      "loss": 0.6955,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5885650518241887e-07,
+      "loss": 0.7822,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.57203183479865e-07,
+      "loss": 0.2918,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5555360990200294e-07,
+      "loss": 0.7334,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5390778509001523e-07,
+      "loss": 0.775,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5226570968363237e-07,
+      "loss": 0.6987,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.506273843211205e-07,
+      "loss": 0.7713,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.489928096392925e-07,
+      "loss": 0.7234,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4736198627350224e-07,
+      "loss": 0.7737,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.457349148576461e-07,
+      "loss": 0.7704,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4411159602416145e-07,
+      "loss": 0.7589,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4249203040402577e-07,
+      "loss": 0.7517,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4087621862676114e-07,
+      "loss": 0.8125,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.392641613204295e-07,
+      "loss": 0.7768,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3765585911162744e-07,
+      "loss": 0.7508,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3605131262550386e-07,
+      "loss": 0.7484,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.344505224857353e-07,
+      "loss": 0.7366,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.328534893145463e-07,
+      "loss": 0.7351,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.312602137326981e-07,
+      "loss": 0.6927,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.296706963594909e-07,
+      "loss": 0.3357,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2808493781276705e-07,
+      "loss": 0.8242,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.265029387089047e-07,
+      "loss": 0.8379,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2492469966282305e-07,
+      "loss": 0.7828,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.233502212879769e-07,
+      "loss": 0.7599,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2177950419636226e-07,
+      "loss": 0.7978,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2021254899851194e-07,
+      "loss": 0.7913,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1864935630349424e-07,
+      "loss": 0.7413,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.170899267189209e-07,
+      "loss": 0.7843,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1553426085093487e-07,
+      "loss": 0.7406,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.139823593042191e-07,
+      "loss": 0.7619,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.124342226819932e-07,
+      "loss": 0.7507,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.108898515860115e-07,
+      "loss": 0.7437,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.093492466165693e-07,
+      "loss": 0.7708,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0781240837249095e-07,
+      "loss": 0.7857,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.06279337451143e-07,
+      "loss": 0.786,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0475003444842553e-07,
+      "loss": 0.766,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.032244999587719e-07,
+      "loss": 0.8478,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.017027345751555e-07,
+      "loss": 0.7666,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0018473888907994e-07,
+      "loss": 0.7488,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.986705134905832e-07,
+      "loss": 0.7192,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9716005896824553e-07,
+      "loss": 0.7708,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9565337590916953e-07,
+      "loss": 0.3194,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9415046489900235e-07,
+      "loss": 0.8169,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9265132652191773e-07,
+      "loss": 0.7611,
+      "step": 4806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.911559613606296e-07,
+      "loss": 0.8061,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.896643699963797e-07,
+      "loss": 0.7837,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.881765530089442e-07,
+      "loss": 0.771,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8669251097663384e-07,
+      "loss": 0.6949,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.852122444762917e-07,
+      "loss": 0.7585,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.837357540832919e-07,
+      "loss": 0.785,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8226304037154206e-07,
+      "loss": 0.3395,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8079410391347873e-07,
+      "loss": 0.7394,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7932894528007625e-07,
+      "loss": 0.31,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7786756504083577e-07,
+      "loss": 0.7919,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.764099637637918e-07,
+      "loss": 0.8035,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.749561420155089e-07,
+      "loss": 0.7972,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7350610036108285e-07,
+      "loss": 0.3483,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.720598393641416e-07,
+      "loss": 0.743,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.706173595868389e-07,
+      "loss": 0.7282,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6917866158986617e-07,
+      "loss": 0.7337,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6774374593243723e-07,
+      "loss": 0.791,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.663126131723015e-07,
+      "loss": 0.7598,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.648852638657362e-07,
+      "loss": 0.3251,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6346169856754534e-07,
+      "loss": 0.7882,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6204191783106516e-07,
+      "loss": 0.8049,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6062592220816195e-07,
+      "loss": 0.8025,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5921371224922554e-07,
+      "loss": 0.7135,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.57805288503179e-07,
+      "loss": 0.7606,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.564006515174722e-07,
+      "loss": 0.3349,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5499980183808284e-07,
+      "loss": 0.7443,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5360274000951645e-07,
+      "loss": 0.7751,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.522094665748087e-07,
+      "loss": 0.781,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.508199820755186e-07,
+      "loss": 0.7332,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49434287051733e-07,
+      "loss": 0.7918,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4805238204207103e-07,
+      "loss": 0.8067,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4667426758367197e-07,
+      "loss": 0.792,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4529994421220506e-07,
+      "loss": 0.7895,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.439294124618652e-07,
+      "loss": 0.7591,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4256267286537404e-07,
+      "loss": 0.8127,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4119972595398e-07,
+      "loss": 0.743,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3984057225745484e-07,
+      "loss": 0.7922,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.384852123040982e-07,
+      "loss": 0.7494,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3713364662073412e-07,
+      "loss": 0.7649,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3578587573271139e-07,
+      "loss": 0.7573,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3444190016390533e-07,
+      "loss": 0.6585,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3310172043671254e-07,
+      "loss": 0.8047,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3176533707205962e-07,
+      "loss": 0.3253,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.304327505893933e-07,
+      "loss": 0.7343,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291039615066848e-07,
+      "loss": 0.8352,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2777897034043206e-07,
+      "loss": 0.8209,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2645777760565313e-07,
+      "loss": 0.7528,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2514038381589387e-07,
+      "loss": 0.786,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2382678948321801e-07,
+      "loss": 0.7343,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2251699511821823e-07,
+      "loss": 0.7859,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2121100123000728e-07,
+      "loss": 0.7626,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1990880832621796e-07,
+      "loss": 0.7162,
+      "step": 4858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.18610416913011e-07,
+      "loss": 0.7819,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1731582749506709e-07,
+      "loss": 0.7684,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1602504057558815e-07,
+      "loss": 0.8004,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1473805665630065e-07,
+      "loss": 0.7057,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1345487623744888e-07,
+      "loss": 0.7134,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1217549981780273e-07,
+      "loss": 0.7308,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1089992789464887e-07,
+      "loss": 0.797,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.096281609638029e-07,
+      "loss": 0.7626,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0836019951959275e-07,
+      "loss": 0.7343,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0709604405487192e-07,
+      "loss": 0.3171,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.058356950610152e-07,
+      "loss": 0.7253,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.04579153027914e-07,
+      "loss": 0.8112,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0332641844398538e-07,
+      "loss": 0.7388,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0207749179615877e-07,
+      "loss": 0.7264,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0083237356989249e-07,
+      "loss": 0.7774,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.995910642491572e-07,
+      "loss": 0.7464,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.983535643164458e-07,
+      "loss": 0.3255,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9711987425277356e-07,
+      "loss": 0.3604,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.958899945376691e-07,
+      "loss": 0.7824,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9466392564918336e-07,
+      "loss": 0.7728,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.934416680638873e-07,
+      "loss": 0.6689,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.922232222568654e-07,
+      "loss": 0.7971,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9100858870172545e-07,
+      "loss": 0.7838,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.89797767870592e-07,
+      "loss": 0.8226,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8859076023410638e-07,
+      "loss": 0.7713,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8738756626142997e-07,
+      "loss": 0.784,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8618818642023863e-07,
+      "loss": 0.3184,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8499262117672945e-07,
+      "loss": 0.7211,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8380087099561295e-07,
+      "loss": 0.7992,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.826129363401208e-07,
+      "loss": 0.7893,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8142881767199804e-07,
+      "loss": 0.7249,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8024851545150769e-07,
+      "loss": 0.785,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7907203013742936e-07,
+      "loss": 0.772,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7789936218706062e-07,
+      "loss": 0.7313,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.767305120562124e-07,
+      "loss": 0.6817,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.755654801992146e-07,
+      "loss": 0.7829,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7440426706890835e-07,
+      "loss": 0.768,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7324687311665812e-07,
+      "loss": 0.717,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7209329879233516e-07,
+      "loss": 0.7578,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7094354454433303e-07,
+      "loss": 0.7538,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6979761081955537e-07,
+      "loss": 0.3368,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6865549806342363e-07,
+      "loss": 0.8147,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.675172067198738e-07,
+      "loss": 0.7866,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6638273723135645e-07,
+      "loss": 0.7403,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6525209003883658e-07,
+      "loss": 0.8088,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6412526558179155e-07,
+      "loss": 0.7749,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6300226429821653e-07,
+      "loss": 0.8228,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6188308662461683e-07,
+      "loss": 0.7617,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076773299601335e-07,
+      "loss": 0.8039,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.596562038459415e-07,
+      "loss": 0.7999,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.585484996064479e-07,
+      "loss": 0.7559,
+      "step": 4909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5744462070809598e-07,
+      "loss": 0.7902,
+      "step": 4910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5634456757995907e-07,
+      "loss": 0.7503,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5524834064962192e-07,
+      "loss": 0.7128,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5415594034318804e-07,
+      "loss": 0.7025,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5306736708526783e-07,
+      "loss": 0.7033,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.519826212989861e-07,
+      "loss": 0.7355,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.509017034059823e-07,
+      "loss": 0.7686,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4982461382640257e-07,
+      "loss": 0.723,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4875135297891085e-07,
+      "loss": 0.7798,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4768192128067905e-07,
+      "loss": 0.7485,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4661631914739238e-07,
+      "loss": 0.2883,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4555454699324733e-07,
+      "loss": 0.8211,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4449660523095155e-07,
+      "loss": 0.6933,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4344249427172164e-07,
+      "loss": 0.7764,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4239221452528873e-07,
+      "loss": 0.8155,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4134576639989294e-07,
+      "loss": 0.7618,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.403031503022867e-07,
+      "loss": 0.7492,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3926436663772912e-07,
+      "loss": 0.7501,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3822941580999393e-07,
+      "loss": 0.7493,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3719829822136267e-07,
+      "loss": 0.7494,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3617101427262913e-07,
+      "loss": 0.7204,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3514756436309173e-07,
+      "loss": 0.7391,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3412794889056445e-07,
+      "loss": 0.7587,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3311216825136919e-07,
+      "loss": 0.7027,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3210022284033563e-07,
+      "loss": 0.7476,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3109211305080361e-07,
+      "loss": 0.76,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.300878392746241e-07,
+      "loss": 0.8239,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2908740190215152e-07,
+      "loss": 0.7927,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2809080132225704e-07,
+      "loss": 0.8161,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2709803792231302e-07,
+      "loss": 0.8001,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.261091120882041e-07,
+      "loss": 0.7626,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.251240242043239e-07,
+      "loss": 0.7971,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.241427746535717e-07,
+      "loss": 0.7882,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2316536381735577e-07,
+      "loss": 0.785,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.221917920755933e-07,
+      "loss": 0.7016,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2122205980671042e-07,
+      "loss": 0.7478,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.202561673876357e-07,
+      "loss": 0.7682,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1929411519380985e-07,
+      "loss": 0.7661,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1833590359918045e-07,
+      "loss": 0.7641,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1738153297619959e-07,
+      "loss": 0.7867,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1643100369582938e-07,
+      "loss": 0.775,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1548431612753652e-07,
+      "loss": 0.736,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1454147063929776e-07,
+      "loss": 0.7938,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1360246759759108e-07,
+      "loss": 0.7484,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1266730736740561e-07,
+      "loss": 0.7901,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1173599031223614e-07,
+      "loss": 0.7854,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.108085167940809e-07,
+      "loss": 0.3306,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0988488717344593e-07,
+      "loss": 0.7696,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0896510180934405e-07,
+      "loss": 0.7504,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0804916105929375e-07,
+      "loss": 0.7294,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0713706527931688e-07,
+      "loss": 0.8569,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0622881482394098e-07,
+      "loss": 0.713,
+      "step": 4961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0532441004620364e-07,
+      "loss": 0.7085,
+      "step": 4962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0442385129764143e-07,
+      "loss": 0.8001,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.035271389282999e-07,
+      "loss": 0.7616,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0263427328672693e-07,
+      "loss": 0.7209,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.017452547199782e-07,
+      "loss": 0.7257,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0086008357361177e-07,
+      "loss": 0.8167,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.997876019169017e-08,
+      "loss": 0.7988,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.910128491678273e-08,
+      "loss": 0.7387,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.822765808995993e-08,
+      "loss": 0.7995,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.735788005079793e-08,
+      "loss": 0.7207,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.649195113737741e-08,
+      "loss": 0.7629,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.562987168628135e-08,
+      "loss": 0.7485,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.477164203259725e-08,
+      "loss": 0.8054,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.391726250991717e-08,
+      "loss": 0.7322,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.306673345033546e-08,
+      "loss": 0.777,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.222005518444876e-08,
+      "loss": 0.7249,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.13772280413605e-08,
+      "loss": 0.7734,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.053825234867308e-08,
+      "loss": 0.8133,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.970312843249563e-08,
+      "loss": 0.722,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.887185661743625e-08,
+      "loss": 0.7131,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.80444372266076e-08,
+      "loss": 0.6872,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.722087058162576e-08,
+      "loss": 0.7338,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.64011570026102e-08,
+      "loss": 0.7765,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.558529680817829e-08,
+      "loss": 0.6945,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.477329031545301e-08,
+      "loss": 0.7438,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.396513784005855e-08,
+      "loss": 0.8326,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.316083969612254e-08,
+      "loss": 0.7852,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.236039619627267e-08,
+      "loss": 0.7987,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.156380765163785e-08,
+      "loss": 0.7505,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.07710743718504e-08,
+      "loss": 0.7574,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.99821966650427e-08,
+      "loss": 0.7513,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.919717483784838e-08,
+      "loss": 0.7429,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.841600919540449e-08,
+      "loss": 0.7551,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.763870004134811e-08,
+      "loss": 0.7962,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.686524767781423e-08,
+      "loss": 0.8102,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.609565240544459e-08,
+      "loss": 0.7842,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.532991452337435e-08,
+      "loss": 0.6822,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.456803432924764e-08,
+      "loss": 0.7857,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.381001211920202e-08,
+      "loss": 0.7735,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.305584818787959e-08,
+      "loss": 0.2975,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.23055428284225e-08,
+      "loss": 0.8184,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.15590963324686e-08,
+      "loss": 0.777,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.081650899016246e-08,
+      "loss": 0.7631,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.007778109014429e-08,
+      "loss": 0.7889,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.934291291955664e-08,
+      "loss": 0.7472,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.861190476403767e-08,
+      "loss": 0.8186,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.788475690773123e-08,
+      "loss": 0.7507,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.716146963327452e-08,
+      "loss": 0.7297,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.644204322180825e-08,
+      "loss": 0.7414,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.572647795297205e-08,
+      "loss": 0.7993,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.501477410490231e-08,
+      "loss": 0.7254,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.430693195423665e-08,
+      "loss": 0.733,
+      "step": 5013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.360295177611164e-08,
+      "loss": 0.7507,
+      "step": 5014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.290283384416063e-08,
+      "loss": 0.69,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.220657843051925e-08,
+      "loss": 0.6482,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.15141858058188e-08,
+      "loss": 0.7989,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.082565623918957e-08,
+      "loss": 0.7436,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.014098999826301e-08,
+      "loss": 0.7246,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9460187349162923e-08,
+      "loss": 0.8035,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8783248556518736e-08,
+      "loss": 0.7044,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.81101738834533e-08,
+      "loss": 0.7645,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.744096359158846e-08,
+      "loss": 0.7546,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6775617941042803e-08,
+      "loss": 0.7251,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6114137190436124e-08,
+      "loss": 0.8109,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.545652159688275e-08,
+      "loss": 0.3128,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.480277141599599e-08,
+      "loss": 0.3179,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4152886901888136e-08,
+      "loss": 0.7618,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.350686830716378e-08,
+      "loss": 0.7386,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.286471588293096e-08,
+      "loss": 0.8555,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2226429878793336e-08,
+      "loss": 0.7072,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1592010542846905e-08,
+      "loss": 0.7293,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.096145812169106e-08,
+      "loss": 0.7725,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0334772860418656e-08,
+      "loss": 0.7391,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.971195500262038e-08,
+      "loss": 0.7562,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9093004790384816e-08,
+      "loss": 0.7714,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8477922464293944e-08,
+      "loss": 0.7752,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7866708263429874e-08,
+      "loss": 0.7652,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7259362425369215e-08,
+      "loss": 0.765,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6655885186185357e-08,
+      "loss": 0.8162,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.605627678044733e-08,
+      "loss": 0.7431,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.546053744122203e-08,
+      "loss": 0.6954,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.48686674000709e-08,
+      "loss": 0.7904,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4280666887051016e-08,
+      "loss": 0.706,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.369653613071845e-08,
+      "loss": 0.7654,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.311627535812157e-08,
+      "loss": 0.7741,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.253988479480664e-08,
+      "loss": 0.7993,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1967364664814433e-08,
+      "loss": 0.7813,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1398715190681394e-08,
+      "loss": 0.3161,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0833936593439594e-08,
+      "loss": 0.7691,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0273029092617877e-08,
+      "loss": 0.8199,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.971599290623851e-08,
+      "loss": 0.7603,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.916282825081941e-08,
+      "loss": 0.6704,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.861353534137524e-08,
+      "loss": 0.7941,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.806811439141189e-08,
+      "loss": 0.75,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.752656561293644e-08,
+      "loss": 0.7302,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6988889216443833e-08,
+      "loss": 0.7689,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.645508541092802e-08,
+      "loss": 0.7229,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.592515440387745e-08,
+      "loss": 0.7937,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5399096401275144e-08,
+      "loss": 0.813,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.487691160759643e-08,
+      "loss": 0.7281,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.43586002258145e-08,
+      "loss": 0.722,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384416245739486e-08,
+      "loss": 0.7774,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.333359850229756e-08,
+      "loss": 0.7394,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.28269085589783e-08,
+      "loss": 0.3565,
+      "step": 5065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2324092824383976e-08,
+      "loss": 0.8013,
+      "step": 5066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1825151493959374e-08,
+      "loss": 0.8072,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1330084761639344e-08,
+      "loss": 0.8375,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.083889281985664e-08,
+      "loss": 0.8029,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0351575859535186e-08,
+      "loss": 0.7469,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.986813407009459e-08,
+      "loss": 0.7313,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.938856763944453e-08,
+      "loss": 0.7537,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.891287675399368e-08,
+      "loss": 0.6968,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8441061598640796e-08,
+      "loss": 0.7399,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.797312235677807e-08,
+      "loss": 0.7588,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7509059210293344e-08,
+      "loss": 0.7578,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7048872339565657e-08,
+      "loss": 0.7136,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6592561923468596e-08,
+      "loss": 0.8839,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.614012813936695e-08,
+      "loss": 0.8139,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5691571163123373e-08,
+      "loss": 0.7477,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5246891169088406e-08,
+      "loss": 0.7804,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4806088330108223e-08,
+      "loss": 0.7648,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.436916281752133e-08,
+      "loss": 0.7855,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.393611480115854e-08,
+      "loss": 0.8041,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3506944449346315e-08,
+      "loss": 0.7355,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3081651928900107e-08,
+      "loss": 0.7697,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2660237405131013e-08,
+      "loss": 0.3127,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.224270104184134e-08,
+      "loss": 0.7501,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.18290430013246e-08,
+      "loss": 0.7401,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1419263444369954e-08,
+      "loss": 0.6784,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1013362530256655e-08,
+      "loss": 0.8113,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0611340416757387e-08,
+      "loss": 0.7443,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021319726013604e-08,
+      "loss": 0.8203,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.981893321515105e-08,
+      "loss": 0.8032,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9428548435052043e-08,
+      "loss": 0.7511,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9042043071577642e-08,
+      "loss": 0.7524,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.865941727496434e-08,
+      "loss": 0.6611,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8280671193935394e-08,
+      "loss": 0.7038,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7905804975708596e-08,
+      "loss": 0.7771,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7534818765994055e-08,
+      "loss": 0.7023,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.716771270899198e-08,
+      "loss": 0.7375,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6804486947397112e-08,
+      "loss": 0.7663,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6445141622392082e-08,
+      "loss": 0.7632,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6089676873654038e-08,
+      "loss": 0.7201,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573809283935135e-08,
+      "loss": 0.7677,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5390389656142478e-08,
+      "loss": 0.7971,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5046567459181537e-08,
+      "loss": 0.7356,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.470662638210718e-08,
+      "loss": 0.7867,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4370566557055932e-08,
+      "loss": 0.8018,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.403838811465219e-08,
+      "loss": 0.7574,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3710091184012674e-08,
+      "loss": 0.7602,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.338567589274642e-08,
+      "loss": 0.743,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3065142366951445e-08,
+      "loss": 0.7897,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2748490731219198e-08,
+      "loss": 0.7972,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2435721108630116e-08,
+      "loss": 0.7482,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2126833620756951e-08,
+      "loss": 0.7411,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1821828387663658e-08,
+      "loss": 0.7908,
+      "step": 5117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1520705527904298e-08,
+      "loss": 0.7593,
+      "step": 5118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1223465158526347e-08,
+      "loss": 0.6964,
+      "step": 5119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.093010739506295e-08,
+      "loss": 0.7895,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0640632351544e-08,
+      "loss": 0.7875,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0355040140486162e-08,
+      "loss": 0.7477,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0073330872898413e-08,
+      "loss": 0.7709,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.79550465828094e-09,
+      "loss": 0.7556,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.52156160462292e-09,
+      "loss": 0.6938,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.25150181840584e-09,
+      "loss": 0.7311,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.985325404601287e-09,
+      "loss": 0.801,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.723032466669834e-09,
+      "loss": 0.7388,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.464623106565484e-09,
+      "loss": 0.7393,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.210097424731223e-09,
+      "loss": 0.7337,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.959455520099025e-09,
+      "loss": 0.7008,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.71269749009318e-09,
+      "loss": 0.7295,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.469823430628076e-09,
+      "loss": 0.7166,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.230833436107087e-09,
+      "loss": 0.8015,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.9957275994270156e-09,
+      "loss": 0.7276,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.764506011970318e-09,
+      "loss": 0.7334,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.537168763613988e-09,
+      "loss": 0.7884,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3137159427217874e-09,
+      "loss": 0.8307,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.094147636150904e-09,
+      "loss": 0.7926,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.878463929245293e-09,
+      "loss": 0.7542,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.666664905840114e-09,
+      "loss": 0.7637,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.458750648262845e-09,
+      "loss": 0.7616,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.254721237327731e-09,
+      "loss": 0.8001,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.054576752341333e-09,
+      "loss": 0.7517,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.85831727109809e-09,
+      "loss": 0.8011,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.665942869884755e-09,
+      "loss": 0.7453,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.477453623474848e-09,
+      "loss": 0.7615,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.292849605134208e-09,
+      "loss": 0.7796,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.112130886619881e-09,
+      "loss": 0.8156,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9352975381734546e-09,
+      "loss": 0.713,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.762349628532169e-09,
+      "loss": 0.6814,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5932872249178077e-09,
+      "loss": 0.7929,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4281103930466906e-09,
+      "loss": 0.724,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2668191971207963e-09,
+      "loss": 0.7693,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1094136998355286e-09,
+      "loss": 0.7263,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9558939623719475e-09,
+      "loss": 0.3246,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8062600444034306e-09,
+      "loss": 0.6921,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6605120040923414e-09,
+      "loss": 0.7262,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.518649898090031e-09,
+      "loss": 0.7716,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3806737815379456e-09,
+      "loss": 0.795,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.246583708067629e-09,
+      "loss": 0.7514,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1163797297996113e-09,
+      "loss": 0.779,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.990061897342299e-09,
+      "loss": 0.7444,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.867630259796416e-09,
+      "loss": 0.712,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7490848647494507e-09,
+      "loss": 0.779,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6344257582812105e-09,
+      "loss": 0.7481,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5236529849582683e-09,
+      "loss": 0.7644,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.416766587838403e-09,
+      "loss": 0.7731,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3137666084672706e-09,
+      "loss": 0.7402,
+      "step": 5169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2146530868806238e-09,
+      "loss": 0.7576,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1194260616043118e-09,
+      "loss": 0.7431,
+      "step": 5171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0280855696520597e-09,
+      "loss": 0.8284,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.4063164652769e-10,
+      "loss": 0.7451,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57064326225121e-10,
+      "loss": 0.3335,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.77383641225038e-10,
+      "loss": 0.8374,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.015896225004426e-10,
+      "loss": 0.7724,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.296822995122132e-10,
+      "loss": 0.8098,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.616617002102142e-10,
+      "loss": 0.7485,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.97527851033297e-10,
+      "loss": 0.7249,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372807769115195e-10,
+      "loss": 0.7237,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.809205012605954e-10,
+      "loss": 0.801,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2844704598966603e-10,
+      "loss": 0.7586,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.798604314946385e-10,
+      "loss": 0.7938,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.351606766592962e-10,
+      "loss": 0.7854,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9434779885973975e-10,
+      "loss": 0.7978,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5742181395994594e-10,
+      "loss": 0.7772,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2438273631176778e-10,
+      "loss": 0.7664,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.523057875826525e-11,
+      "loss": 0.7901,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.996535263148473e-11,
+      "loss": 0.7921,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.858706775023869e-11,
+      "loss": 0.3183,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.109573242565667e-11,
+      "loss": 0.7171,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749135345563424e-11,
+      "loss": 0.7629,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.773936128163684e-12,
+      "loss": 0.7432,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9434842213339466e-12,
+      "loss": 0.7108,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8103,
+      "step": 5195
+    },
+    {
+      "epoch": 1.0,
+      "step": 5195,
+      "total_flos": 1.076204591054848e+16,
+      "train_loss": 0.7809832206417209,
+      "train_runtime": 66435.0901,
+      "train_samples_per_second": 10.011,
+      "train_steps_per_second": 0.078
+    }
+  ],
+  "max_steps": 5195,
+  "num_train_epochs": 1,
+  "total_flos": 1.076204591054848e+16,
+  "trial_name": null,
+  "trial_params": null
+}