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": 1.0,
+  "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-06,
+      "loss": 1.6297,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.4484,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.6281,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.5574,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 1.2868,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.4748,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.2506,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.4785,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 1.2121,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 1.1479,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 1.1839,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.1181,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.0435,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.9962,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.9838,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 1.0796,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 0.9551,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 1.0173,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 0.9469,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 1.0091,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 0.9466,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 1.1317,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 1.1031,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.754,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 0.92,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.8023,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 0.832,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 1.0581,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 0.9684,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.7854,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 0.8506,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 0.9066,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 0.9055,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 0.9265,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 0.7942,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 0.9997,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 0.9542,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 1.0002,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 0.9401,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 0.9016,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-05,
+      "loss": 0.8643,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 0.8543,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-05,
+      "loss": 0.7877,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 0.916,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 0.7763,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 0.8649,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-05,
+      "loss": 0.793,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 0.8761,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-05,
+      "loss": 0.7461,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 0.9819,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-05,
+      "loss": 0.8797,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.7542,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871795e-05,
+      "loss": 0.8841,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 0.8301,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282052e-05,
+      "loss": 0.8719,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 0.9982,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692307e-05,
+      "loss": 0.8723,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 0.875,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-05,
+      "loss": 0.8244,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 0.8389,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512821e-05,
+      "loss": 0.9709,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 0.7531,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923078e-05,
+      "loss": 0.8293,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 1.0195,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.9679,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 0.9131,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-05,
+      "loss": 0.907,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 0.7997,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 0.7985,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 0.8188,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564103e-05,
+      "loss": 0.8193,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 0.7156,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.35897435897436e-05,
+      "loss": 0.7996,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 0.7638,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 0.8887,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 0.9541,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-05,
+      "loss": 0.7068,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 0.7489,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010128205128205129,
+      "loss": 0.851,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010256410256410256,
+      "loss": 0.705,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010384615384615386,
+      "loss": 0.8284,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010512820512820514,
+      "loss": 0.8272,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010641025641025641,
+      "loss": 0.8196,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 0.8469,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010897435897435896,
+      "loss": 0.7608,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011025641025641027,
+      "loss": 0.9948,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011153846153846154,
+      "loss": 0.7965,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011282051282051283,
+      "loss": 0.7901,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001141025641025641,
+      "loss": 0.771,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 0.8502,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 0.8663,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011794871794871796,
+      "loss": 0.8495,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011923076923076923,
+      "loss": 0.8688,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012051282051282052,
+      "loss": 0.8218,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012179487179487179,
+      "loss": 0.8626,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.7577,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012435897435897437,
+      "loss": 0.9641,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012564102564102564,
+      "loss": 0.7296,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012692307692307693,
+      "loss": 0.7463,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012820512820512823,
+      "loss": 0.8193,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001294871794871795,
+      "loss": 0.9028,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 0.7936,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013205128205128204,
+      "loss": 0.7972,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.9391,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 0.8472,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 0.8361,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013717948717948718,
+      "loss": 0.8496,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 1.0547,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013974358974358974,
+      "loss": 0.7914,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014102564102564104,
+      "loss": 0.8812,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001423076923076923,
+      "loss": 0.8518,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001435897435897436,
+      "loss": 0.6487,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014487179487179488,
+      "loss": 0.9674,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 0.8281,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014743589743589745,
+      "loss": 0.8074,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014871794871794872,
+      "loss": 0.8391,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.7215,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015128205128205128,
+      "loss": 0.7933,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015256410256410255,
+      "loss": 0.7289,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 0.8526,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015512820512820515,
+      "loss": 0.8198,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015641025641025642,
+      "loss": 0.7924,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001576923076923077,
+      "loss": 0.9047,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015897435897435896,
+      "loss": 0.8425,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016025641025641028,
+      "loss": 0.776,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 0.943,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016282051282051282,
+      "loss": 0.7544,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001641025641025641,
+      "loss": 0.712,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001653846153846154,
+      "loss": 0.8091,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.7927,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016794871794871796,
+      "loss": 0.9225,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 0.8473,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017051282051282053,
+      "loss": 0.7513,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001717948717948718,
+      "loss": 0.8972,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 0.7181,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017435897435897436,
+      "loss": 0.837,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017564102564102566,
+      "loss": 0.6799,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 0.7069,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017820512820512823,
+      "loss": 0.8106,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001794871794871795,
+      "loss": 0.7168,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018076923076923077,
+      "loss": 0.8998,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018205128205128207,
+      "loss": 0.7734,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.828,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 0.7973,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001858974358974359,
+      "loss": 0.8079,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001871794871794872,
+      "loss": 0.808,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018846153846153847,
+      "loss": 0.7701,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018974358974358974,
+      "loss": 0.8302,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019102564102564104,
+      "loss": 0.7738,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 0.7211,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001935897435897436,
+      "loss": 0.7633,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019487179487179487,
+      "loss": 0.9393,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 0.7373,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019743589743589744,
+      "loss": 0.7904,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019871794871794874,
+      "loss": 0.8683,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7015,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999999805651578,
+      "loss": 0.8021,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999992226063873,
+      "loss": 0.826,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999982508646545,
+      "loss": 0.8715,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999968904267578,
+      "loss": 0.6881,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999995141293225,
+      "loss": 0.993,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999993003464737,
+      "loss": 0.858,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999904769421242,
+      "loss": 0.7707,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999987561726369,
+      "loss": 0.8114,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999984257818604,
+      "loss": 0.7747,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999805652201141,
+      "loss": 0.9996,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999976483932334,
+      "loss": 0.739,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999720139568508,
+      "loss": 0.8184,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999967155295401,
+      "loss": 0.9284,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999961907949874,
+      "loss": 0.8058,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999956271922309,
+      "loss": 0.7581,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999502472148968,
+      "loss": 0.7947,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999943833829979,
+      "loss": 0.8828,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999370317700489,
+      "loss": 0.8302,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199992984103775,
+      "loss": 0.8914,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999222616358777,
+      "loss": 0.9461,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999142935673775,
+      "loss": 0.7342,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999059368353472,
+      "loss": 0.7799,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999897191443035,
+      "loss": 0.6892,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998880573938397,
+      "loss": 0.7845,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999878534691312,
+      "loss": 0.8204,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998686233391535,
+      "loss": 0.7448,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998583233412162,
+      "loss": 0.8848,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998476347015044,
+      "loss": 0.9236,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999836557424172,
+      "loss": 0.7027,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998250915135253,
+      "loss": 0.8312,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998132369740206,
+      "loss": 0.6795,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998009938102657,
+      "loss": 0.8393,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199978836202702,
+      "loss": 0.93,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997753416291933,
+      "loss": 0.8295,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997619326218464,
+      "loss": 0.8693,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999748135010191,
+      "loss": 0.8562,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999733948799591,
+      "loss": 0.7864,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997193739955597,
+      "loss": 0.9613,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999704410603763,
+      "loss": 0.7205,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996890586300166,
+      "loss": 0.9045,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999673318080288,
+      "loss": 0.7792,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996571889606955,
+      "loss": 0.9793,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996406712775083,
+      "loss": 0.7553,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999623765037147,
+      "loss": 0.6762,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996064702461827,
+      "loss": 0.7099,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995887869113382,
+      "loss": 0.7819,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995707150394866,
+      "loss": 0.9283,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995522546376526,
+      "loss": 0.8152,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995334057130117,
+      "loss": 0.7164,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995141682728904,
+      "loss": 0.8623,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999494542324766,
+      "loss": 0.8978,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994745278762674,
+      "loss": 0.7365,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994541249351738,
+      "loss": 0.8315,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999433333509416,
+      "loss": 0.8009,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994121536070757,
+      "loss": 0.8289,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999390585236385,
+      "loss": 0.8269,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999368628405728,
+      "loss": 0.8055,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993462831236386,
+      "loss": 0.8165,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999323549398803,
+      "loss": 0.9265,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993004272400575,
+      "loss": 0.635,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992769166563893,
+      "loss": 0.7927,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992530176569373,
+      "loss": 0.8129,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999228730250991,
+      "loss": 0.765,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992040544479901,
+      "loss": 0.8688,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991789902575272,
+      "loss": 0.8305,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991535376893437,
+      "loss": 0.9591,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999127696753333,
+      "loss": 0.7372,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199910146745954,
+      "loss": 0.9431,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990748498181595,
+      "loss": 0.7688,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999047843839538,
+      "loss": 0.8384,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999020449534172,
+      "loss": 0.7762,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989926669127103,
+      "loss": 0.79,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989644959859514,
+      "loss": 0.7832,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989359367648458,
+      "loss": 0.7905,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998906989260494,
+      "loss": 0.7556,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988776534841476,
+      "loss": 0.7573,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988479294472095,
+      "loss": 0.7995,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988178171612337,
+      "loss": 0.8653,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987873166379245,
+      "loss": 0.7951,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998756427889137,
+      "loss": 0.6732,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998725150926878,
+      "loss": 0.8646,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998693485763305,
+      "loss": 0.8233,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986614324107253,
+      "loss": 0.9275,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986289908815988,
+      "loss": 0.757,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985961611885348,
+      "loss": 0.8508,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985629433442946,
+      "loss": 0.7091,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985293373617894,
+      "loss": 0.6704,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998495343254082,
+      "loss": 0.7709,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984609610343857,
+      "loss": 0.8049,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984261907160652,
+      "loss": 0.9077,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983910323126346,
+      "loss": 0.8116,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983554858377608,
+      "loss": 0.8456,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983195513052603,
+      "loss": 0.8481,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998283228729101,
+      "loss": 0.9169,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982465181234008,
+      "loss": 0.7988,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982094195024292,
+      "loss": 0.8747,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981719328806066,
+      "loss": 0.825,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981340582725038,
+      "loss": 0.753,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980957956928423,
+      "loss": 0.7242,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980571451564948,
+      "loss": 0.8847,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998018106678485,
+      "loss": 0.7019,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979786802739864,
+      "loss": 0.7336,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979388659583245,
+      "loss": 0.8084,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978986637469746,
+      "loss": 0.8703,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997858073655563,
+      "loss": 0.8491,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978170956998676,
+      "loss": 0.8657,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997775729895816,
+      "loss": 0.7505,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997733976259487,
+      "loss": 0.7766,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976918348071102,
+      "loss": 0.8392,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976493055550655,
+      "loss": 0.7376,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976063885198843,
+      "loss": 0.9021,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997563083718248,
+      "loss": 0.7578,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975193911669892,
+      "loss": 0.7406,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974753108830914,
+      "loss": 0.8098,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997430842883688,
+      "loss": 0.7481,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973859871860633,
+      "loss": 0.8243,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973407438076532,
+      "loss": 0.7813,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972951127660436,
+      "loss": 0.9713,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972490940789708,
+      "loss": 0.7826,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972026877643223,
+      "loss": 0.8383,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997155893840136,
+      "loss": 0.8049,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971087123246006,
+      "loss": 0.9562,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970611432360556,
+      "loss": 0.7093,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970131865929908,
+      "loss": 0.8324,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969648424140465,
+      "loss": 0.8579,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969161107180145,
+      "loss": 0.8299,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968669915238364,
+      "loss": 0.7765,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968174848506042,
+      "loss": 0.8862,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996767590717562,
+      "loss": 0.8579,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967173091441022,
+      "loss": 0.7877,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966666401497705,
+      "loss": 0.7151,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966155837542607,
+      "loss": 0.6568,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965641399774186,
+      "loss": 0.7318,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965123088392406,
+      "loss": 0.6973,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964600903598726,
+      "loss": 0.7585,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964074845596122,
+      "loss": 0.743,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996354491458907,
+      "loss": 0.7998,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996301111078356,
+      "loss": 0.713,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962473434387065,
+      "loss": 0.8131,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996193188560859,
+      "loss": 0.7233,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961386464658625,
+      "loss": 0.8401,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960837171749182,
+      "loss": 0.8298,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960284007093762,
+      "loss": 0.9068,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959726970907382,
+      "loss": 0.8848,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959166063406562,
+      "loss": 0.8681,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995860128480932,
+      "loss": 0.8309,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958032635335188,
+      "loss": 0.8576,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957460115205194,
+      "loss": 0.8164,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995688372464188,
+      "loss": 0.702,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956303463869283,
+      "loss": 0.7179,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995571933311295,
+      "loss": 0.8653,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955131332599933,
+      "loss": 0.8997,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995453946255878,
+      "loss": 0.8468,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953943723219553,
+      "loss": 1.0044,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953344114813818,
+      "loss": 0.6722,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952740637574633,
+      "loss": 0.7131,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995213329173657,
+      "loss": 0.7686,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951522077535708,
+      "loss": 0.9117,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950906995209615,
+      "loss": 0.7866,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995028804499738,
+      "loss": 0.8067,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949665227139582,
+      "loss": 0.8292,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994903854187831,
+      "loss": 0.8036,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948407989457155,
+      "loss": 0.7279,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994777357012121,
+      "loss": 0.7613,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994713528411707,
+      "loss": 0.7305,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994649313169284,
+      "loss": 0.8868,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945847113098114,
+      "loss": 0.7186,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945197228584004,
+      "loss": 0.8426,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994454347840312,
+      "loss": 0.8031,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943885862809565,
+      "loss": 0.9657,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943224382058957,
+      "loss": 0.7518,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942559036408413,
+      "loss": 0.7319,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941889826116546,
+      "loss": 0.75,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941216751443483,
+      "loss": 0.8699,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940539812650837,
+      "loss": 0.655,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993985901000174,
+      "loss": 0.7995,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939174343760812,
+      "loss": 0.8468,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938485814194182,
+      "loss": 0.8471,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993779342156948,
+      "loss": 0.9027,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993709716615584,
+      "loss": 0.7688,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993639704822389,
+      "loss": 0.8842,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935693068045765,
+      "loss": 0.8571,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934985225895097,
+      "loss": 0.6636,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934273522047028,
+      "loss": 0.7686,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933557956778192,
+      "loss": 0.8386,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932838530366725,
+      "loss": 0.8835,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993211524309227,
+      "loss": 0.7277,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931388095235963,
+      "loss": 0.8421,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930657087080444,
+      "loss": 0.6635,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929922218909856,
+      "loss": 0.6834,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929183491009837,
+      "loss": 0.8149,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992844090366753,
+      "loss": 0.7942,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992769445717158,
+      "loss": 0.8419,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992694415181212,
+      "loss": 0.8293,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199261899878808,
+      "loss": 1.0193,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925431965670754,
+      "loss": 0.7401,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924670085476627,
+      "loss": 0.7309,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923904347594556,
+      "loss": 0.87,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992313475232219,
+      "loss": 0.684,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922361299958653,
+      "loss": 0.8132,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921583990804598,
+      "loss": 0.7605,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920802825162154,
+      "loss": 0.942,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920017803334958,
+      "loss": 0.9277,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991922892562815,
+      "loss": 0.8982,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918436192348362,
+      "loss": 0.75,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991763960380373,
+      "loss": 0.873,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916839160303876,
+      "loss": 0.7497,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916034862159943,
+      "loss": 0.7875,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991522670968455,
+      "loss": 0.7541,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914414703191825,
+      "loss": 0.8444,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991359884299739,
+      "loss": 0.7759,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912779129418375,
+      "loss": 0.8107,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911955562773392,
+      "loss": 0.8365,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911128143382568,
+      "loss": 0.7309,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910296871567508,
+      "loss": 0.6889,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909461747651328,
+      "loss": 0.7625,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908622771958642,
+      "loss": 0.6952,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907779944815552,
+      "loss": 0.707,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906933266549665,
+      "loss": 0.8004,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906082737490083,
+      "loss": 0.8524,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905228357967403,
+      "loss": 0.62,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990437012831372,
+      "loss": 0.7933,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903508048862626,
+      "loss": 0.7203,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019902642119949202,
+      "loss": 0.8511,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990177234191004,
+      "loss": 0.7835,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990089871508322,
+      "loss": 0.7839,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990002123980831,
+      "loss": 0.8146,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019899139916426388,
+      "loss": 0.7838,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898254745280021,
+      "loss": 0.8259,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897365726713275,
+      "loss": 0.7273,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896472861071702,
+      "loss": 0.8495,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989557614870236,
+      "loss": 0.7587,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019894675589953798,
+      "loss": 0.8675,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989377118517606,
+      "loss": 0.8831,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892862934720685,
+      "loss": 0.8134,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891950838940708,
+      "loss": 0.7385,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891034898190656,
+      "loss": 0.7531,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019890115112826555,
+      "loss": 0.7978,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988919148320592,
+      "loss": 0.713,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888264009687765,
+      "loss": 0.8019,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887332692632595,
+      "loss": 0.8743,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988639753240241,
+      "loss": 0.7507,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885458529360703,
+      "loss": 0.8531,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019884515683872464,
+      "loss": 0.7244,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988356899630417,
+      "loss": 0.6924,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882618467023801,
+      "loss": 0.7122,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988166409640082,
+      "loss": 0.7032,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988070588480619,
+      "loss": 0.7364,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879743832612365,
+      "loss": 0.87,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987877794019329,
+      "loss": 0.7237,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877808207924408,
+      "loss": 0.9419,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019876834636182645,
+      "loss": 0.8497,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987585722534643,
+      "loss": 0.6606,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874875975795677,
+      "loss": 0.9122,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873890887911797,
+      "loss": 0.8925,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872901962077687,
+      "loss": 0.8445,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871909198677744,
+      "loss": 0.7926,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019870912598097848,
+      "loss": 0.6889,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986991216072538,
+      "loss": 0.7717,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198689078869492,
+      "loss": 0.7709,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867899777159664,
+      "loss": 0.7635,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866887831748633,
+      "loss": 0.8528,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865872051109436,
+      "loss": 0.7942,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986485243563691,
+      "loss": 0.7172,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863828985727374,
+      "loss": 0.9073,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862801701778636,
+      "loss": 0.7932,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019861770584190006,
+      "loss": 0.8195,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860735633362274,
+      "loss": 0.8304,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019859696849697716,
+      "loss": 0.8135,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019858654233600106,
+      "loss": 0.7647,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857607785474712,
+      "loss": 0.7365,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985655750572828,
+      "loss": 0.7142,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985550339476905,
+      "loss": 0.7993,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854445453006754,
+      "loss": 0.7217,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985338368085261,
+      "loss": 0.936,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852318078719322,
+      "loss": 0.7793,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985124864702109,
+      "loss": 0.7214,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019850175386173596,
+      "loss": 0.8117,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984909829659402,
+      "loss": 0.7835,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848017378701016,
+      "loss": 0.8521,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846932632914733,
+      "loss": 0.7526,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845844059656812,
+      "loss": 0.8688,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984475165935038,
+      "loss": 0.7647,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843655432420044,
+      "loss": 0.7534,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019842555379291906,
+      "loss": 0.7041,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001984145150039355,
+      "loss": 0.6297,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001984034379615406,
+      "loss": 0.7073,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983923226700399,
+      "loss": 0.8965,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019838116913375387,
+      "loss": 0.7149,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019836997735701785,
+      "loss": 0.8467,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983587473441821,
+      "loss": 0.9074,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019834747909961163,
+      "loss": 0.7831,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833617262768644,
+      "loss": 0.7406,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019832482793280128,
+      "loss": 0.7477,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831344501936577,
+      "loss": 0.8412,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830202389180447,
+      "loss": 0.7459,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982905645545567,
+      "loss": 0.8218,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827906701207665,
+      "loss": 0.7925,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826753126883344,
+      "loss": 0.8107,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019825595732931092,
+      "loss": 0.6939,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019824434519800788,
+      "loss": 0.7775,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019823269487943788,
+      "loss": 0.6628,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982210063781294,
+      "loss": 0.7993,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019820927969862572,
+      "loss": 0.7289,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019819751484548496,
+      "loss": 0.8135,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818571182328003,
+      "loss": 0.8143,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019817387063659881,
+      "loss": 0.7847,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816199129004387,
+      "loss": 0.8778,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981500737882327,
+      "loss": 0.7873,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981381181357976,
+      "loss": 0.8812,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981261243373857,
+      "loss": 0.7604,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019811409239765894,
+      "loss": 0.7726,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810202232129409,
+      "loss": 0.8882,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019808991411298274,
+      "loss": 0.7022,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807776777743137,
+      "loss": 0.7854,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019806558331936117,
+      "loss": 0.7491,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019805336074350818,
+      "loss": 0.891,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019804110005462332,
+      "loss": 0.7502,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019802880125747229,
+      "loss": 0.7313,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801646435683553,
+      "loss": 0.6659,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019800408935750845,
+      "loss": 0.6969,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019799167626430108,
+      "loss": 0.7605,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019797922508203844,
+      "loss": 0.7539,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019796673581556017,
+      "loss": 0.8295,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795420846972087,
+      "loss": 0.8244,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794164304938984,
+      "loss": 0.699,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979290395594513,
+      "loss": 0.7117,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019791639800480408,
+      "loss": 0.7728,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197903718390362,
+      "loss": 0.7016,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019789100072105354,
+      "loss": 0.7201,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197878245001822,
+      "loss": 0.7872,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019786545123762556,
+      "loss": 0.8608,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785261943343701,
+      "loss": 0.8065,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019783974959424413,
+      "loss": 0.7685,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019782684172504932,
+      "loss": 0.7738,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978138958308699,
+      "loss": 0.7622,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019780091191673785,
+      "loss": 0.7352,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019778788998769996,
+      "loss": 0.8892,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019777483004881783,
+      "loss": 0.7353,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019776173210516782,
+      "loss": 0.8456,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774859616184106,
+      "loss": 0.8266,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977354222239435,
+      "loss": 0.7843,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977222102965957,
+      "loss": 0.8648,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019770896038493317,
+      "loss": 0.8916,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976956724941061,
+      "loss": 0.7888,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019768234662927942,
+      "loss": 0.8022,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019766898279563287,
+      "loss": 0.92,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019765558099836097,
+      "loss": 0.8745,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976421412426729,
+      "loss": 0.727,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019762866353379267,
+      "loss": 0.8429,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019761514787695902,
+      "loss": 0.7781,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019760159427742547,
+      "loss": 0.8233,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975880027404602,
+      "loss": 0.8151,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757437327134626,
+      "loss": 0.8813,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019756070587538137,
+      "loss": 0.7246,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754700055787797,
+      "loss": 0.7492,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019753325732416331,
+      "loss": 0.7606,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019751947617957928,
+      "loss": 0.736,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975056571294827,
+      "loss": 0.7675,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749180017924483,
+      "loss": 0.947,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019747790533425192,
+      "loss": 0.8413,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019746397259990485,
+      "loss": 0.6752,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974500019816192,
+      "loss": 0.8267,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974359934848253,
+      "loss": 0.8302,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019742194711496822,
+      "loss": 0.8271,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019740786287750776,
+      "loss": 0.8658,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019739374077791842,
+      "loss": 0.7841,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019737958082168935,
+      "loss": 0.7965,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736538301432454,
+      "loss": 0.8797,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735114736134265,
+      "loss": 0.7498,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197336873868277,
+      "loss": 0.7249,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019732256254067565,
+      "loss": 0.7682,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019730821338410137,
+      "loss": 0.8975,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019729382640413162,
+      "loss": 0.8316,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972794016063586,
+      "loss": 0.726,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972649389963892,
+      "loss": 0.7271,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972504385798449,
+      "loss": 0.6547,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972359003623621,
+      "loss": 0.7835,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019722132434959165,
+      "loss": 0.84,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019720671054719927,
+      "loss": 0.8569,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719205896086522,
+      "loss": 0.6775,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717736959628458,
+      "loss": 0.7109,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019716264245916708,
+      "loss": 0.7651,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019714787755523708,
+      "loss": 0.8127,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713307489023366,
+      "loss": 0.9465,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019711823446991056,
+      "loss": 0.8944,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019710335630003623,
+      "loss": 0.8156,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019708844038639371,
+      "loss": 0.9965,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019707348673478085,
+      "loss": 0.7479,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019705849535101,
+      "loss": 0.7339,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704346624090834,
+      "loss": 0.7474,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019702839941031756,
+      "loss": 0.8125,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019701329486509418,
+      "loss": 0.752,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019699815261110923,
+      "loss": 0.8033,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019698297265424846,
+      "loss": 0.71,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019696775500041227,
+      "loss": 0.7025,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019695249965551574,
+      "loss": 0.7503,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019693720662548859,
+      "loss": 0.6424,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969218759162751,
+      "loss": 0.7169,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019690650753383432,
+      "loss": 0.7534,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968911014841399,
+      "loss": 0.6441,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019687565777318006,
+      "loss": 0.7634,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019686017640695782,
+      "loss": 0.7356,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019684465739149064,
+      "loss": 0.8101,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968291007328108,
+      "loss": 0.8511,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681350643696506,
+      "loss": 0.8525,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967978745100149,
+      "loss": 0.7854,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019678220495803638,
+      "loss": 0.8825,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019676649778712025,
+      "loss": 0.7101,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019675075300337177,
+      "loss": 0.9572,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019673497061291096,
+      "loss": 0.7079,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019671915062187233,
+      "loss": 0.8679,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967032930364051,
+      "loss": 0.7167,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019668739786267306,
+      "loss": 0.7412,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667146510685455,
+      "loss": 0.6712,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019665549477514266,
+      "loss": 0.7524,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196639486873745,
+      "loss": 0.8402,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019662344140888372,
+      "loss": 0.7418,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019660735838679573,
+      "loss": 0.8489,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965912378137324,
+      "loss": 0.9197,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019657507969595973,
+      "loss": 0.8263,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965588840397584,
+      "loss": 0.677,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019654265085142355,
+      "loss": 0.7806,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196526380137265,
+      "loss": 0.869,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965100719036071,
+      "loss": 0.7917,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019649372615678884,
+      "loss": 0.7966,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019647734290316368,
+      "loss": 0.7144,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019646092214909984,
+      "loss": 0.9039,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019644446390097996,
+      "loss": 0.7365,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019642796816520135,
+      "loss": 0.891,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019641143494817582,
+      "loss": 0.8058,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019639486425632977,
+      "loss": 0.7633,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963782560961042,
+      "loss": 0.8272,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019636161047395466,
+      "loss": 0.6997,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019634492739635116,
+      "loss": 0.8607,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019632820686977848,
+      "loss": 0.7747,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963114489007358,
+      "loss": 0.7156,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019629465349573684,
+      "loss": 0.6994,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019627782066130998,
+      "loss": 0.7721,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019626095040399806,
+      "loss": 0.7928,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962440427303585,
+      "loss": 0.6326,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019622709764696323,
+      "loss": 0.8513,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019621011516039882,
+      "loss": 0.8369,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019619309527726624,
+      "loss": 0.7234,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019617603800418113,
+      "loss": 0.7865,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019615894334777353,
+      "loss": 0.69,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019614181131468814,
+      "loss": 0.7864,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019612464191158408,
+      "loss": 0.7817,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019610743514513506,
+      "loss": 0.8276,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960901910220293,
+      "loss": 0.7652,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019607290954896952,
+      "loss": 0.8967,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000196055590732673,
+      "loss": 0.7832,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019603823457987153,
+      "loss": 0.7606,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960208410973113,
+      "loss": 0.9111,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019600341029175322,
+      "loss": 0.732,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959859421699725,
+      "loss": 0.765,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019596843673875899,
+      "loss": 0.7327,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019595089400491696,
+      "loss": 0.8134,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019593331397526524,
+      "loss": 0.9664,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019591569665663712,
+      "loss": 0.6916,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019589804205588044,
+      "loss": 0.8362,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958803501798574,
+      "loss": 0.8721,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019586262103544485,
+      "loss": 0.6955,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019584485462953403,
+      "loss": 0.9256,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019582705096903068,
+      "loss": 0.7474,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019580921006085504,
+      "loss": 0.7825,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019579133191194182,
+      "loss": 0.7129,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019577341652924016,
+      "loss": 0.7708,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019575546391971375,
+      "loss": 0.8093,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957374740903407,
+      "loss": 0.7375,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019571944704811358,
+      "loss": 0.6769,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957013828000395,
+      "loss": 0.8034,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019568328135313994,
+      "loss": 0.8051,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019566514271445086,
+      "loss": 0.7752,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019564696689102274,
+      "loss": 0.7719,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019562875388992043,
+      "loss": 0.8619,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019561050371822325,
+      "loss": 0.8704,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019559221638302502,
+      "loss": 0.7884,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955738918914339,
+      "loss": 0.7734,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955555302505727,
+      "loss": 0.791,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955371314675784,
+      "loss": 0.759,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019551869554960264,
+      "loss": 0.7639,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955002225038113,
+      "loss": 0.8174,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954817123373849,
+      "loss": 0.8313,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019546316505751824,
+      "loss": 0.7389,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019544458067142055,
+      "loss": 0.7829,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954259591863156,
+      "loss": 0.8461,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954073006094414,
+      "loss": 0.7842,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953886049480506,
+      "loss": 0.6971,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953698722094101,
+      "loss": 0.7036,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019535110240080118,
+      "loss": 0.74,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019533229552951974,
+      "loss": 0.8028,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019531345160287583,
+      "loss": 0.7262,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019529457062819407,
+      "loss": 0.8499,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019527565261281347,
+      "loss": 0.7949,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019525669756408735,
+      "loss": 0.8194,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952377054893835,
+      "loss": 0.9064,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019521867639608412,
+      "loss": 0.7595,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951996102915857,
+      "loss": 0.6507,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951805071832992,
+      "loss": 0.7953,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951613670786499,
+      "loss": 0.6461,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514218998507754,
+      "loss": 0.7495,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019512297591003618,
+      "loss": 0.8207,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951037248609943,
+      "loss": 0.7958,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019508443684543468,
+      "loss": 0.889,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019506511187085456,
+      "loss": 0.8776,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019504574994476541,
+      "loss": 0.7601,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019502635107469324,
+      "loss": 0.7522,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950069152681783,
+      "loss": 0.8027,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019498744253277522,
+      "loss": 0.9056,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019496793287605295,
+      "loss": 0.7777,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949483863055949,
+      "loss": 0.8204,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019492880282899876,
+      "loss": 0.7302,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949091824538765,
+      "loss": 0.8466,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019488952518785456,
+      "loss": 0.8595,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019486983103857362,
+      "loss": 0.718,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019485010001368878,
+      "loss": 0.8432,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019483033212086936,
+      "loss": 0.6433,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019481052736779913,
+      "loss": 0.7805,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947906857621761,
+      "loss": 0.7503,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019477080731171266,
+      "loss": 0.9397,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947508920241355,
+      "loss": 0.8308,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019473093990718567,
+      "loss": 0.7598,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019471095096861842,
+      "loss": 0.8489,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019469092521620342,
+      "loss": 0.8505,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946708626577246,
+      "loss": 0.7159,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019465076330098024,
+      "loss": 0.9113,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946306271537829,
+      "loss": 1.041,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019461045422395943,
+      "loss": 0.8293,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019459024451935094,
+      "loss": 0.7199,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019456999804781297,
+      "loss": 0.7398,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019454971481721515,
+      "loss": 0.6859,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019452939483544158,
+      "loss": 0.8812,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945090381103906,
+      "loss": 0.6233,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019448864464997472,
+      "loss": 0.7785,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019446821446212085,
+      "loss": 0.793,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019444774755477017,
+      "loss": 0.7664,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019442724393587806,
+      "loss": 0.7884,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019440670361341422,
+      "loss": 0.7936,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019438612659536264,
+      "loss": 0.7863,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943655128897215,
+      "loss": 0.6624,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943448625045033,
+      "loss": 0.8641,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019432417544773477,
+      "loss": 0.786,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019430345172745695,
+      "loss": 0.9169,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019428269135172502,
+      "loss": 0.6894,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019426189432860853,
+      "loss": 0.797,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019424106066619117,
+      "loss": 0.8546,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019422019037257092,
+      "loss": 0.6745,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019419928345586004,
+      "loss": 0.7962,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019417833992418495,
+      "loss": 0.7501,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019415735978568636,
+      "loss": 0.849,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019413634304851914,
+      "loss": 0.7453,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019411528972085247,
+      "loss": 0.7748,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019409419981086966,
+      "loss": 0.8441,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940730733267684,
+      "loss": 0.7711,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019405191027676034,
+      "loss": 0.7812,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019403071066907162,
+      "loss": 0.7979,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940094745119424,
+      "loss": 0.7974,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019398820181362712,
+      "loss": 0.7793,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939668925823944,
+      "loss": 0.6499,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019394554682652704,
+      "loss": 0.8717,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019392416455432218,
+      "loss": 0.7704,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939027457740909,
+      "loss": 0.715,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019388129049415867,
+      "loss": 0.7897,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019385979872286514,
+      "loss": 0.7365,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019383827046856405,
+      "loss": 0.8702,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019381670573962337,
+      "loss": 0.6994,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937951045444252,
+      "loss": 0.7649,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937734668913659,
+      "loss": 0.8827,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193751792788856,
+      "loss": 0.6804,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019373008224532007,
+      "loss": 0.8011,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193708335269197,
+      "loss": 0.8207,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936865518689397,
+      "loss": 0.6985,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019366473205301534,
+      "loss": 0.7865,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019364287582990526,
+      "loss": 0.7921,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019362098320810484,
+      "loss": 0.7899,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019359905419612372,
+      "loss": 0.7554,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935770888024856,
+      "loss": 0.6748,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019355508703572838,
+      "loss": 0.7748,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935330489044041,
+      "loss": 0.8554,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019351097441707883,
+      "loss": 0.7788,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019348886358233296,
+      "loss": 0.9207,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934667164087608,
+      "loss": 0.7979,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019344453290497096,
+      "loss": 0.7703,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019342231307958608,
+      "loss": 0.7695,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019340005694124294,
+      "loss": 0.7556,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019337776449859242,
+      "loss": 0.8061,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019335543576029952,
+      "loss": 0.8865,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019333307073504335,
+      "loss": 0.6388,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019331066943151711,
+      "loss": 0.8242,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019328823185842814,
+      "loss": 0.7458,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019326575802449788,
+      "loss": 0.8633,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932432479384618,
+      "loss": 0.8022,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019322070160906948,
+      "loss": 0.7762,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019319811904508464,
+      "loss": 0.7763,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019317550025528505,
+      "loss": 0.668,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019315284524846257,
+      "loss": 0.692,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001931301540334231,
+      "loss": 0.7508,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019310742661898667,
+      "loss": 0.6553,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019308466301398732,
+      "loss": 0.8276,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019306186322727323,
+      "loss": 0.7968,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019303902726770659,
+      "loss": 0.847,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930161551441637,
+      "loss": 0.7467,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929932468655348,
+      "loss": 0.7059,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019297030244072434,
+      "loss": 0.8866,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019294732187865072,
+      "loss": 0.8054,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929243051882464,
+      "loss": 0.6921,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929012523784579,
+      "loss": 0.8322,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928781634582458,
+      "loss": 0.8743,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019285503843658467,
+      "loss": 0.8524,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019283187732246313,
+      "loss": 0.7837,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019280868012488384,
+      "loss": 0.8069,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019278544685286347,
+      "loss": 0.8067,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019276217751543274,
+      "loss": 0.8514,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927388721216363,
+      "loss": 0.7304,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019271553068053295,
+      "loss": 0.7593,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019269215320119547,
+      "loss": 0.7799,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019266873969271054,
+      "loss": 0.8011,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019264529016417892,
+      "loss": 0.7303,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019262180462471538,
+      "loss": 0.8683,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925982830834487,
+      "loss": 0.7432,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925747255495216,
+      "loss": 0.6916,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019255113203209083,
+      "loss": 0.802,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019252750254032712,
+      "loss": 0.7426,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019250383708341518,
+      "loss": 0.6792,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019248013567055368,
+      "loss": 0.7038,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019245639831095527,
+      "loss": 0.803,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019243262501384667,
+      "loss": 0.84,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019240881578846839,
+      "loss": 0.7991,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019238497064407505,
+      "loss": 0.8506,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019236108958993519,
+      "loss": 0.7362,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019233717263533124,
+      "loss": 0.8108,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923132197895597,
+      "loss": 0.7246,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019228923106193098,
+      "loss": 0.7995,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019226520646176937,
+      "loss": 0.906,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019224114599841318,
+      "loss": 0.7949,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019221704968121466,
+      "loss": 0.8269,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921929175195399,
+      "loss": 0.7258,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019216874952276912,
+      "loss": 0.8173,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019214454570029623,
+      "loss": 0.8062,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019212030606152923,
+      "loss": 0.6638,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019209603061588998,
+      "loss": 0.6272,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019207171937281423,
+      "loss": 0.7924,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019204737234175176,
+      "loss": 0.7131,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019202298953216614,
+      "loss": 0.8011,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019199857095353489,
+      "loss": 0.8432,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019197411661534946,
+      "loss": 0.905,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019194962652711514,
+      "loss": 0.7601,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019192510069835118,
+      "loss": 0.7303,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019190053913859068,
+      "loss": 0.7575,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918759418573806,
+      "loss": 0.7884,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019185130886428188,
+      "loss": 0.6833,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918266401688693,
+      "loss": 0.8123,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019180193578073146,
+      "loss": 0.7817,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917771957094709,
+      "loss": 0.7623,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019175241996470402,
+      "loss": 0.7543,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191727608556061,
+      "loss": 0.7604,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019170276149318606,
+      "loss": 0.9063,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916778787857371,
+      "loss": 0.7103,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019165296044338596,
+      "loss": 0.7232,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019162800647581836,
+      "loss": 0.7303,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916030168927338,
+      "loss": 0.737,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019157799170384568,
+      "loss": 0.8794,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019155293091888118,
+      "loss": 0.7893,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915278345475813,
+      "loss": 0.8706,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019150270259970107,
+      "loss": 0.6746,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019147753508500905,
+      "loss": 0.7517,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019145233201328784,
+      "loss": 0.7531,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914270933943338,
+      "loss": 0.7329,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019140181923795705,
+      "loss": 0.7919,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019137650955398163,
+      "loss": 0.8244,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019135116435224533,
+      "loss": 0.7734,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019132578364259973,
+      "loss": 0.7676,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019130036743491023,
+      "loss": 0.7571,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019127491573905602,
+      "loss": 0.8139,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019124942856493012,
+      "loss": 0.7009,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912239059224393,
+      "loss": 0.7695,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019119834782150416,
+      "loss": 0.846,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019117275427205904,
+      "loss": 0.7073,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191147125284052,
+      "loss": 0.932,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019112146086744505,
+      "loss": 0.8136,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910957610322138,
+      "loss": 0.7296,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019107002578834777,
+      "loss": 0.6647,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019104425514585005,
+      "loss": 0.774,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019101844911473772,
+      "loss": 0.8247,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019099260770504144,
+      "loss": 0.7825,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909667309268057,
+      "loss": 0.9125,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019094081879008875,
+      "loss": 0.866,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909148713049625,
+      "loss": 0.867,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908888884815127,
+      "loss": 0.7458,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019086287032983877,
+      "loss": 0.7829,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019083681686005388,
+      "loss": 0.858,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019081072808228497,
+      "loss": 0.7039,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907846040066726,
+      "loss": 0.7897,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907584446433712,
+      "loss": 0.9323,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019073225000254876,
+      "loss": 0.7119,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070602009438708,
+      "loss": 0.7243,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019067975492908165,
+      "loss": 0.7489,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019065345451684164,
+      "loss": 0.7555,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019062711886788994,
+      "loss": 0.6747,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019060074799246316,
+      "loss": 0.8004,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019057434190081153,
+      "loss": 0.8101,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019054790060319906,
+      "loss": 0.6367,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019052142410990337,
+      "loss": 0.745,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001904949124312158,
+      "loss": 0.8217,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019046836557744135,
+      "loss": 0.7621,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001904417835588987,
+      "loss": 0.7672,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019041516638592022,
+      "loss": 0.7987,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001903885140688519,
+      "loss": 0.7981,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001903618266180534,
+      "loss": 0.8136,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019033510404389806,
+      "loss": 0.7585,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019030834635677286,
+      "loss": 0.8412,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019028155356707842,
+      "loss": 0.7821,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019025472568522903,
+      "loss": 0.7425,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019022786272165261,
+      "loss": 0.8079,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001902009646867907,
+      "loss": 0.8427,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019017403159109843,
+      "loss": 0.7092,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019014706344504468,
+      "loss": 0.7452,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019012006025911186,
+      "loss": 0.8134,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019009302204379603,
+      "loss": 0.7285,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019006594880960681,
+      "loss": 0.7928,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019003884056706757,
+      "loss": 0.8831,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900116973267151,
+      "loss": 0.829,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899845190991,
+      "loss": 0.7908,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018995730589478626,
+      "loss": 0.7658,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018993005772435163,
+      "loss": 0.7635,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018990277459838734,
+      "loss": 0.8443,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898754565274983,
+      "loss": 0.7853,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018984810352230293,
+      "loss": 0.7698,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018982071559343327,
+      "loss": 0.8538,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018979329275153493,
+      "loss": 0.7387,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018976583500726705,
+      "loss": 0.8511,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897383423713024,
+      "loss": 0.6428,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018971081485432727,
+      "loss": 0.7001,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896832524670415,
+      "loss": 0.7287,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018965565522015857,
+      "loss": 0.7497,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018962802312440534,
+      "loss": 0.808,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018960035619052242,
+      "loss": 0.7792,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018957265442926376,
+      "loss": 0.7349,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018954491785139703,
+      "loss": 0.8888,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895171464677033,
+      "loss": 0.7827,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018948934028897724,
+      "loss": 0.775,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018946149932602707,
+      "loss": 0.8069,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894336235896744,
+      "loss": 0.7256,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940571309075446,
+      "loss": 0.7423,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018937776784011602,
+      "loss": 0.8655,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018934978784862128,
+      "loss": 0.7647,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018932177312714593,
+      "loss": 0.6844,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018929372368657932,
+      "loss": 0.8392,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018926563953782405,
+      "loss": 0.7793,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018923752069179646,
+      "loss": 0.7279,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018920936715942615,
+      "loss": 0.8213,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018918117895165638,
+      "loss": 0.7571,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018915295607944374,
+      "loss": 0.7554,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018912469855375847,
+      "loss": 0.7734,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018909640638558415,
+      "loss": 0.7934,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890680795859178,
+      "loss": 0.826,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018903971816577002,
+      "loss": 0.786,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018901132213616477,
+      "loss": 0.7047,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018898289150813953,
+      "loss": 0.8625,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018895442629274516,
+      "loss": 0.7176,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018892592650104602,
+      "loss": 0.882,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018889739214411987,
+      "loss": 0.6801,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018886882323305794,
+      "loss": 0.6783,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888402197789649,
+      "loss": 0.8042,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888115817929588,
+      "loss": 0.8827,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001887829092861711,
+      "loss": 0.8729,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018875420226974676,
+      "loss": 0.7726,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001887254607548441,
+      "loss": 0.7786,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018869668475263483,
+      "loss": 0.6802,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886678742743041,
+      "loss": 0.7345,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886390293310505,
+      "loss": 0.7715,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018861014993408587,
+      "loss": 0.7843,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018858123609463565,
+      "loss": 0.8386,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885522878239385,
+      "loss": 0.7877,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885233051332465,
+      "loss": 0.9075,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018849428803382519,
+      "loss": 0.6598,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018846523653695337,
+      "loss": 0.7783,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884361506539233,
+      "loss": 0.7282,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018840703039604058,
+      "loss": 0.7734,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883778757746241,
+      "loss": 0.7854,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018834868680100622,
+      "loss": 0.746,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018831946348653263,
+      "loss": 0.772,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018829020584256226,
+      "loss": 0.7167,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018826091388046755,
+      "loss": 0.7832,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018823158761163412,
+      "loss": 0.7244,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018820222704746104,
+      "loss": 0.7851,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018817283219936065,
+      "loss": 0.8134,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018814340307875868,
+      "loss": 0.8205,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018811393969709404,
+      "loss": 0.8693,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018808444206581916,
+      "loss": 0.7298,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880549101963996,
+      "loss": 0.7199,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018802534410031435,
+      "loss": 0.8756,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018799574378905565,
+      "loss": 0.7903,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018796610927412902,
+      "loss": 0.7299,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879364405670533,
+      "loss": 0.7729,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018790673767936066,
+      "loss": 0.7913,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018787700062259651,
+      "loss": 0.9145,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018784722940831956,
+      "loss": 0.714,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018781742404810173,
+      "loss": 0.8542,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018778758455352835,
+      "loss": 0.8106,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018775771093619786,
+      "loss": 0.7572,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018772780320772215,
+      "loss": 0.8526,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018769786137972608,
+      "loss": 0.7497,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018766788546384814,
+      "loss": 0.7137,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018763787547173975,
+      "loss": 0.9437,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018760783141506574,
+      "loss": 0.6511,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018757775330550413,
+      "loss": 0.7757,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018754764115474617,
+      "loss": 0.8112,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875174949744964,
+      "loss": 0.6985,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874873147764725,
+      "loss": 0.8786,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874571005724055,
+      "loss": 0.7281,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018742685237403946,
+      "loss": 0.8106,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018739657019313183,
+      "loss": 0.8216,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018736625404145314,
+      "loss": 0.7482,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018733590393078724,
+      "loss": 0.8551,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873055198729311,
+      "loss": 0.7069,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872751018796949,
+      "loss": 0.76,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018724464996290203,
+      "loss": 0.7001,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018721416413438905,
+      "loss": 0.6774,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871836444060057,
+      "loss": 0.6841,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871530907896149,
+      "loss": 0.7901,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018712250329709277,
+      "loss": 0.7854,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018709188194032854,
+      "loss": 0.72,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018706122673122465,
+      "loss": 0.8632,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018703053768169667,
+      "loss": 0.843,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018699981480367335,
+      "loss": 0.8243,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018696905810909657,
+      "loss": 0.821,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018693826760992138,
+      "loss": 0.8026,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001869074433181159,
+      "loss": 0.7276,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868765852456615,
+      "loss": 0.7824,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018684569340455254,
+      "loss": 0.7286,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018681476780679668,
+      "loss": 0.7258,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018678380846441455,
+      "loss": 0.7891,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867528153894399,
+      "loss": 0.8299,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867217885939197,
+      "loss": 0.8668,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018669072808991395,
+      "loss": 0.8357,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018665963388949576,
+      "loss": 0.7345,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018662850600475138,
+      "loss": 0.7865,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018659734444778013,
+      "loss": 0.6683,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865661492306943,
+      "loss": 0.8251,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018653492036561952,
+      "loss": 0.7753,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018650365786469425,
+      "loss": 0.7465,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001864723617400702,
+      "loss": 0.8025,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186441032003912,
+      "loss": 0.7389,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018640966866839744,
+      "loss": 0.7128,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018637827174571738,
+      "loss": 0.8641,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001863468412480757,
+      "loss": 0.6816,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018631537718768928,
+      "loss": 0.6787,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001862838795767882,
+      "loss": 0.8639,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018625234842761538,
+      "loss": 0.8381,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018622078375242696,
+      "loss": 0.5625,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186189185563492,
+      "loss": 0.6699,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018615755387309255,
+      "loss": 0.7807,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018612588869352387,
+      "loss": 0.7461,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186094190037094,
+      "loss": 0.8248,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001860624579161242,
+      "loss": 0.7166,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018603069234294858,
+      "loss": 0.7738,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018599889332991437,
+      "loss": 0.7855,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018596706088938172,
+      "loss": 0.6956,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859351950337238,
+      "loss": 0.8816,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018590329577532674,
+      "loss": 0.8258,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018587136312658972,
+      "loss": 0.8144,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018583939709992486,
+      "loss": 0.7572,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018580739770775723,
+      "loss": 0.7276,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001857753649625249,
+      "loss": 0.6876,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018574329887667893,
+      "loss": 0.7412,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018571119946268324,
+      "loss": 0.6694,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018567906673301482,
+      "loss": 0.7068,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018564690070016357,
+      "loss": 0.8133,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018561470137663227,
+      "loss": 0.7833,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018558246877493678,
+      "loss": 0.8845,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855502029076057,
+      "loss": 0.813,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855179037871808,
+      "loss": 0.8491,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018548557142621653,
+      "loss": 0.8248,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018545320583728044,
+      "loss": 0.7339,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001854208070329529,
+      "loss": 0.7775,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018538837502582727,
+      "loss": 0.7674,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853559098285097,
+      "loss": 0.8693,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853234114536194,
+      "loss": 0.7478,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001852908799137883,
+      "loss": 0.7357,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018525831522166135,
+      "loss": 0.708,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001852257173898963,
+      "loss": 0.9083,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851930864311639,
+      "loss": 0.7096,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018516042235814767,
+      "loss": 0.8338,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018512772518354398,
+      "loss": 0.8133,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018509499492006218,
+      "loss": 0.8966,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850622315804244,
+      "loss": 0.8241,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018502943517736565,
+      "loss": 0.843,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018499660572363378,
+      "loss": 0.8166,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849637432319895,
+      "loss": 0.7892,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018493084771520633,
+      "loss": 0.7991,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018489791918607074,
+      "loss": 0.6969,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018486495765738186,
+      "loss": 0.7591,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018483196314195174,
+      "loss": 0.8405,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018479893565260524,
+      "loss": 0.7226,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847658752021801,
+      "loss": 0.684,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847327818035268,
+      "loss": 0.8865,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846996554695086,
+      "loss": 0.7814,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846664962130016,
+      "loss": 0.6821,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846333040468947,
+      "loss": 0.7462,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846000789840896,
+      "loss": 0.7263,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001845668210375008,
+      "loss": 0.7626,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018453353022005557,
+      "loss": 0.7636,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018450020654469388,
+      "loss": 0.8747,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001844668500243686,
+      "loss": 0.8788,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018443346067204526,
+      "loss": 0.8752,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018440003850070223,
+      "loss": 0.7925,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018436658352333056,
+      "loss": 0.7773,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018433309575293413,
+      "loss": 0.8049,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842995752025295,
+      "loss": 0.6463,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018426602188514608,
+      "loss": 0.8378,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018423243581382589,
+      "loss": 0.8325,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018419881700162364,
+      "loss": 0.7819,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000184165165461607,
+      "loss": 0.7501,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018413148120685616,
+      "loss": 0.761,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018409776425046403,
+      "loss": 0.7319,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001840640146055364,
+      "loss": 0.7389,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018403023228519153,
+      "loss": 0.8395,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018399641730256055,
+      "loss": 0.7971,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839625696707873,
+      "loss": 0.822,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018392868940302813,
+      "loss": 0.7742,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001838947765124523,
+      "loss": 0.7462,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001838608310122416,
+      "loss": 0.797,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001838268529155905,
+      "loss": 0.7241,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018379284223570623,
+      "loss": 0.7261,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018375879898580862,
+      "loss": 0.8089,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018372472317913018,
+      "loss": 0.7701,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018369061482891605,
+      "loss": 0.7158,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018365647394842405,
+      "loss": 0.8046,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018362230055092465,
+      "loss": 0.7239,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018358809464970088,
+      "loss": 0.8191,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018355385625804855,
+      "loss": 0.7733,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018351958538927597,
+      "loss": 0.7274,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001834852820567041,
+      "loss": 0.77,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018345094627366656,
+      "loss": 0.7501,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001834165780535096,
+      "loss": 0.7289,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018338217740959198,
+      "loss": 0.6534,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833477443552851,
+      "loss": 0.677,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018331327890397308,
+      "loss": 0.787,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018327878106905244,
+      "loss": 0.8886,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018324425086393243,
+      "loss": 0.6629,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832096883020348,
+      "loss": 0.7811,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831750933967939,
+      "loss": 0.6594,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018314046616165673,
+      "loss": 0.7861,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831058066100827,
+      "loss": 0.734,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018307111475554394,
+      "loss": 0.6788,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018303639061152503,
+      "loss": 0.6394,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018300163419152315,
+      "loss": 0.7318,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018296684550904796,
+      "loss": 0.8001,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018293202457762177,
+      "loss": 0.8105,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018289717141077932,
+      "loss": 0.8226,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018286228602206798,
+      "loss": 0.801,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018282736842504757,
+      "loss": 0.7408,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018279241863329042,
+      "loss": 0.7768,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018275743666038144,
+      "loss": 0.7403,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000182722422519918,
+      "loss": 0.7569,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018268737622550996,
+      "loss": 0.7284,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018265229779077975,
+      "loss": 0.6917,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018261718722936224,
+      "loss": 0.7456,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018258204455490472,
+      "loss": 0.756,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018254686978106717,
+      "loss": 0.672,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001825116629215218,
+      "loss": 0.7362,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018247642398995346,
+      "loss": 0.8534,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018244115300005938,
+      "loss": 0.7877,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001824058499655493,
+      "loss": 0.7134,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018237051490014542,
+      "loss": 0.7963,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001823351478175823,
+      "loss": 0.724,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018229974873160712,
+      "loss": 0.7676,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018226431765597928,
+      "loss": 0.8057,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822288546044708,
+      "loss": 0.8073,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018219335959086604,
+      "loss": 0.8035,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821578326289618,
+      "loss": 0.6625,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018212227373256727,
+      "loss": 0.7243,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018208668291550413,
+      "loss": 0.8371,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001820510601916064,
+      "loss": 0.8882,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018201540557472048,
+      "loss": 0.7654,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018197971907870524,
+      "loss": 0.7399,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018194400071743194,
+      "loss": 0.7682,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018190825050478412,
+      "loss": 0.8205,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018187246845465787,
+      "loss": 0.6534,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018183665458096145,
+      "loss": 0.7019,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818008088976157,
+      "loss": 0.6832,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018176493141855367,
+      "loss": 0.8496,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018172902215772083,
+      "loss": 0.6795,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181693081129075,
+      "loss": 0.7198,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018165710834658634,
+      "loss": 0.6463,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018162110382423738,
+      "loss": 0.8829,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018158506757602293,
+      "loss": 0.7516,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018154899961595017,
+      "loss": 0.7242,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018151289995803864,
+      "loss": 0.7661,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018147676861632014,
+      "loss": 0.8272,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018144060560483876,
+      "loss": 0.7063,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018140441093765104,
+      "loss": 0.7574,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018136818462882569,
+      "loss": 0.7611,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018133192669244372,
+      "loss": 0.8119,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018129563714259855,
+      "loss": 0.6992,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018125931599339575,
+      "loss": 0.7434,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018122296325895327,
+      "loss": 0.7973,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001811865789534013,
+      "loss": 0.8749,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001811501630908823,
+      "loss": 0.8531,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018111371568555102,
+      "loss": 0.8953,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018107723675157443,
+      "loss": 0.907,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018104072630313175,
+      "loss": 0.7652,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018100418435441454,
+      "loss": 0.7009,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809676109196265,
+      "loss": 0.8702,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809310060129836,
+      "loss": 0.6989,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018089436964871406,
+      "loss": 0.7798,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018085770184105834,
+      "loss": 0.8549,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808210026042691,
+      "loss": 0.7166,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001807842719526112,
+      "loss": 0.8207,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018074750990036174,
+      "loss": 0.8089,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018071071646181,
+      "loss": 0.7939,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001806738916512575,
+      "loss": 0.6499,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018063703548301788,
+      "loss": 0.6748,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018060014797141705,
+      "loss": 0.8682,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018056322913079305,
+      "loss": 0.7963,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018052627897549613,
+      "loss": 0.6904,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018048929751988873,
+      "loss": 0.8398,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018045228477834538,
+      "loss": 0.7024,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018041524076525285,
+      "loss": 0.7105,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018037816549500998,
+      "loss": 0.7024,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018034105898202785,
+      "loss": 0.7901,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018030392124072963,
+      "loss": 0.9073,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018026675228555067,
+      "loss": 0.6383,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018022955213093837,
+      "loss": 0.7466,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018019232079135238,
+      "loss": 0.8171,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018015505828126438,
+      "loss": 0.675,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018011776461515817,
+      "loss": 0.6913,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018008043980752967,
+      "loss": 0.8642,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018004308387288695,
+      "loss": 0.6569,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018000569682575016,
+      "loss": 0.7993,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017996827868065144,
+      "loss": 0.8297,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001799308294521352,
+      "loss": 0.7312,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798933491547578,
+      "loss": 0.7377,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798558378030877,
+      "loss": 0.712,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017981829541170545,
+      "loss": 0.7567,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017978072199520368,
+      "loss": 0.7385,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017974311756818707,
+      "loss": 0.7834,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797054821452723,
+      "loss": 0.642,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017966781574108812,
+      "loss": 0.6542,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017963011837027543,
+      "loss": 0.704,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000179592390047487,
+      "loss": 0.7608,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017955463078738775,
+      "loss": 0.6951,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017951684060465458,
+      "loss": 0.7339,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794790195139764,
+      "loss": 0.7093,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017944116753005417,
+      "loss": 0.739,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794032846676008,
+      "loss": 0.8627,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017936537094134124,
+      "loss": 0.7674,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017932742636601252,
+      "loss": 0.7483,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792894509563635,
+      "loss": 0.7305,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792514447271551,
+      "loss": 0.6948,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017921340769316024,
+      "loss": 0.833,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791753398691638,
+      "loss": 0.7086,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017913724126996262,
+      "loss": 0.7542,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017909911191036547,
+      "loss": 0.7701,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017906095180519318,
+      "loss": 0.7595,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001790227609692784,
+      "loss": 0.8394,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017898453941746583,
+      "loss": 0.8916,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017894628716461205,
+      "loss": 0.6938,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001789080042255856,
+      "loss": 0.7018,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017886969061526692,
+      "loss": 0.7139,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788313463485484,
+      "loss": 0.6891,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001787929714403343,
+      "loss": 0.7339,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001787545659055409,
+      "loss": 0.7538,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017871612975909625,
+      "loss": 0.8346,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017867766301594038,
+      "loss": 0.8416,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017863916569102519,
+      "loss": 0.6814,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017860063779931448,
+      "loss": 0.6815,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017856207935578388,
+      "loss": 0.7873,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017852349037542097,
+      "loss": 0.8136,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017848487087322517,
+      "loss": 0.7727,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017844622086420775,
+      "loss": 0.8732,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017840754036339184,
+      "loss": 0.7016,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783688293858124,
+      "loss": 0.8139,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017833008794651634,
+      "loss": 0.6998,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001782913160605623,
+      "loss": 0.7393,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017825251374302075,
+      "loss": 0.7401,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017821368100897405,
+      "loss": 0.6923,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781748178735164,
+      "loss": 0.8039,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781359243517537,
+      "loss": 0.8966,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017809700045880384,
+      "loss": 0.7598,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017805804620979633,
+      "loss": 0.732,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001780190616198726,
+      "loss": 0.786,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017798004670418587,
+      "loss": 0.7295,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017794100147790104,
+      "loss": 0.902,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017790192595619491,
+      "loss": 0.7299,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017786282015425605,
+      "loss": 0.6385,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017782368408728467,
+      "loss": 0.786,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017778451777049288,
+      "loss": 0.7373,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017774532121910452,
+      "loss": 0.7652,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017770609444835517,
+      "loss": 0.8159,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776668374734921,
+      "loss": 0.6943,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017762755030977442,
+      "loss": 0.7391,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017758823297247295,
+      "loss": 0.7754,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017754888547687013,
+      "loss": 0.7638,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017750950783826028,
+      "loss": 0.7474,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017747010007194934,
+      "loss": 0.7357,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017743066219325497,
+      "loss": 0.7246,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017739119421750658,
+      "loss": 0.6989,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773516961600452,
+      "loss": 0.6846,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017731216803622366,
+      "loss": 0.7791,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001772726098614064,
+      "loss": 0.6999,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017723302165096954,
+      "loss": 0.8258,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017719340342030088,
+      "loss": 0.6734,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017715375518479994,
+      "loss": 0.7635,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017711407695987781,
+      "loss": 0.8132,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017707436876095736,
+      "loss": 0.8449,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177034630603473,
+      "loss": 0.7094,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769948625028708,
+      "loss": 0.654,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017695506447460856,
+      "loss": 0.7575,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017691523653415563,
+      "loss": 0.7458,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017687537869699296,
+      "loss": 0.736,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017683549097861318,
+      "loss": 0.7939,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017679557339452058,
+      "loss": 0.7213,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017675562596023095,
+      "loss": 0.6699,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001767156486912717,
+      "loss": 0.6183,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017667564160318193,
+      "loss": 0.7992,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017663560471151225,
+      "loss": 0.8121,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017659553803182486,
+      "loss": 0.6539,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017655544157969355,
+      "loss": 0.7186,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017651531537070373,
+      "loss": 0.859,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017647515942045226,
+      "loss": 0.7575,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017643497374454767,
+      "loss": 0.8022,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017639475835860998,
+      "loss": 0.7844,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017635451327827086,
+      "loss": 0.6953,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017631423851917336,
+      "loss": 0.75,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017627393409697217,
+      "loss": 0.8357,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017623360002733352,
+      "loss": 0.8755,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001761932363259351,
+      "loss": 0.7279,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001761528430084662,
+      "loss": 0.8258,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017611242009062754,
+      "loss": 0.697,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760719675881314,
+      "loss": 0.7038,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760314855167015,
+      "loss": 0.7417,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017599097389207313,
+      "loss": 0.6723,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017595043272999305,
+      "loss": 0.8291,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017590986204621945,
+      "loss": 0.7934,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175869261856522,
+      "loss": 0.8417,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001758286321766819,
+      "loss": 0.77,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757879730224918,
+      "loss": 0.7477,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017574728440975573,
+      "loss": 0.7675,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757065663542893,
+      "loss": 0.7401,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001756658188719194,
+      "loss": 0.6769,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017562504197848452,
+      "loss": 0.7965,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017558423568983446,
+      "loss": 0.7138,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017554340002183053,
+      "loss": 0.693,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017550253499034541,
+      "loss": 0.6724,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017546164061126322,
+      "loss": 0.7484,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001754207169004795,
+      "loss": 0.7941,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017537976387390107,
+      "loss": 0.7229,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017533878154744634,
+      "loss": 0.9392,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017529776993704496,
+      "loss": 0.842,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017525672905863808,
+      "loss": 0.7045,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017521565892817805,
+      "loss": 0.5801,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751745595616288,
+      "loss": 0.8,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751334309749655,
+      "loss": 0.7463,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017509227318417465,
+      "loss": 0.7931,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017505108620525424,
+      "loss": 0.8562,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017500987005421344,
+      "loss": 0.6629,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017496862474707285,
+      "loss": 0.6962,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017492735029986443,
+      "loss": 0.7787,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017488604672863143,
+      "loss": 0.7633,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017484471404942834,
+      "loss": 0.7352,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017480335227832114,
+      "loss": 0.7157,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017476196143138695,
+      "loss": 0.6333,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017472054152471432,
+      "loss": 0.7794,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000174679092574403,
+      "loss": 0.7331,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017463761459656402,
+      "loss": 0.8073,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017459610760731984,
+      "loss": 0.6927,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017455457162280402,
+      "loss": 0.8017,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001745130066591615,
+      "loss": 0.7755,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017447141273254846,
+      "loss": 0.7317,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017442978985913226,
+      "loss": 0.7689,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017438813805509168,
+      "loss": 0.8207,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017434645733661658,
+      "loss": 0.7306,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017430474771990814,
+      "loss": 0.8437,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017426300922117875,
+      "loss": 0.8053,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017422124185665204,
+      "loss": 0.7297,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017417944564256286,
+      "loss": 0.6624,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001741376205951572,
+      "loss": 0.8663,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017409576673069245,
+      "loss": 0.8621,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017405388406543695,
+      "loss": 0.6992,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017401197261567047,
+      "loss": 0.8339,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017397003239768378,
+      "loss": 0.7295,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017392806342777893,
+      "loss": 0.7493,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017388606572226912,
+      "loss": 0.6318,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017384403929747878,
+      "loss": 0.7936,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017380198416974335,
+      "loss": 0.7497,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001737599003554096,
+      "loss": 0.7444,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017371778787083538,
+      "loss": 0.7275,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736756467323896,
+      "loss": 0.7076,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736334769564525,
+      "loss": 0.8334,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017359127855941524,
+      "loss": 0.764,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017354905155768027,
+      "loss": 0.7923,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017350679596766106,
+      "loss": 0.8903,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017346451180578224,
+      "loss": 0.9133,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001734221990884795,
+      "loss": 0.8275,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733798578321997,
+      "loss": 0.6648,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733374880534007,
+      "loss": 0.7001,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001732950897685516,
+      "loss": 0.7143,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017325266299413237,
+      "loss": 0.7093,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017321020774663422,
+      "loss": 0.7558,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017316772404255934,
+      "loss": 0.6837,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017312521189842106,
+      "loss": 0.7434,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017308267133074368,
+      "loss": 0.7388,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017304010235606257,
+      "loss": 0.7027,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017299750499092417,
+      "loss": 0.8034,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017295487925188596,
+      "loss": 0.6539,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001729122251555164,
+      "loss": 0.7171,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017286954271839507,
+      "loss": 0.8226,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017282683195711242,
+      "loss": 0.7182,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017278409288827,
+      "loss": 0.8196,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727413255284804,
+      "loss": 0.7115,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017269852989436712,
+      "loss": 0.7299,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726557060025647,
+      "loss": 0.7674,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726128538697186,
+      "loss": 0.8374,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725699735124854,
+      "loss": 0.8418,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017252706494753252,
+      "loss": 0.7422,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017248412819153837,
+      "loss": 0.6911,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001724411632611923,
+      "loss": 0.837,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017239817017319472,
+      "loss": 0.6455,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017235514894425687,
+      "loss": 0.7606,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017231209959110095,
+      "loss": 0.9077,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017226902213046016,
+      "loss": 0.704,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001722259165790785,
+      "loss": 0.7271,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000172182782953711,
+      "loss": 0.8465,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017213962127112352,
+      "loss": 0.7515,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000172096431548093,
+      "loss": 0.8476,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000172053213801407,
+      "loss": 0.632,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001720099680478642,
+      "loss": 0.725,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001719666943042741,
+      "loss": 0.7504,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171923392587457,
+      "loss": 0.7591,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017188006291424422,
+      "loss": 0.7085,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017183670530147778,
+      "loss": 0.7328,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717933197660107,
+      "loss": 0.7206,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717499063247068,
+      "loss": 0.7241,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017170646499444078,
+      "loss": 0.5918,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017166299579209806,
+      "loss": 0.7906,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017161949873457507,
+      "loss": 0.6855,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017157597383877891,
+      "loss": 0.737,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017153242112162765,
+      "loss": 0.7882,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017148884060004998,
+      "loss": 0.7958,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714452322909856,
+      "loss": 0.7504,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714015962113849,
+      "loss": 0.7785,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017135793237820908,
+      "loss": 0.6472,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017131424080843016,
+      "loss": 0.7734,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017127052151903085,
+      "loss": 0.667,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017122677452700478,
+      "loss": 0.7353,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017118299984935622,
+      "loss": 0.6735,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017113919750310025,
+      "loss": 0.8661,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017109536750526273,
+      "loss": 0.703,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017105150987288022,
+      "loss": 0.7674,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017100762462300004,
+      "loss": 0.7773,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017096371177268026,
+      "loss": 0.7318,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017091977133898968,
+      "loss": 0.816,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017087580333900778,
+      "loss": 0.8447,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017083180778982478,
+      "loss": 0.763,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017078778470854165,
+      "loss": 0.9402,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017074373411226997,
+      "loss": 0.7599,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001706996560181321,
+      "loss": 0.7668,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001706555504432611,
+      "loss": 0.6729,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017061141740480057,
+      "loss": 0.6804,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017056725691990494,
+      "loss": 0.7986,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017052306900573925,
+      "loss": 0.792,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704788536794792,
+      "loss": 0.6633,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704346109583111,
+      "loss": 0.7553,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017039034085943207,
+      "loss": 0.7647,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017034604340004968,
+      "loss": 0.785,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017030171859738218,
+      "loss": 0.6672,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017025736646865853,
+      "loss": 0.7861,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017021298703111826,
+      "loss": 0.6703,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017016858030201152,
+      "loss": 0.7756,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017012414629859906,
+      "loss": 0.7468,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017007968503815222,
+      "loss": 0.7622,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017003519653795298,
+      "loss": 0.6783,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016999068081529383,
+      "loss": 0.8126,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016994613788747794,
+      "loss": 0.7788,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016990156777181903,
+      "loss": 0.8734,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016985697048564128,
+      "loss": 0.6755,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016981234604627958,
+      "loss": 0.8903,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016976769447107928,
+      "loss": 0.7154,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001697230157773963,
+      "loss": 0.7818,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016967830998259714,
+      "loss": 0.8379,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016963357710405878,
+      "loss": 0.7744,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016958881715916876,
+      "loss": 0.7454,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016954403016532511,
+      "loss": 0.7595,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016949921613993642,
+      "loss": 0.6911,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001694543751004217,
+      "loss": 0.704,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001694095070642106,
+      "loss": 0.7648,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016936461204874318,
+      "loss": 0.7301,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016931969007146988,
+      "loss": 0.7884,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016927474114985188,
+      "loss": 0.7805,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016922976530136058,
+      "loss": 0.7792,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169184762543478,
+      "loss": 0.7316,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016913973289369656,
+      "loss": 0.7872,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690946763695191,
+      "loss": 0.7552,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169049592988459,
+      "loss": 0.7314,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016900448276804003,
+      "loss": 0.7535,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016895934572579637,
+      "loss": 0.6922,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016891418187927265,
+      "loss": 0.7984,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001688689912460239,
+      "loss": 0.7082,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016882377384361562,
+      "loss": 0.7396,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687785296896236,
+      "loss": 0.8843,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687332588016342,
+      "loss": 0.6688,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016868796119724397,
+      "loss": 0.7511,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016864263689406002,
+      "loss": 0.826,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685972859096997,
+      "loss": 0.779,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016855190826179086,
+      "loss": 0.725,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016850650396797162,
+      "loss": 0.769,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016846107304589045,
+      "loss": 0.668,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016841561551320626,
+      "loss": 0.7146,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016837013138758824,
+      "loss": 0.7878,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016832462068671586,
+      "loss": 0.6792,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682790834282791,
+      "loss": 0.6968,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016823351962997807,
+      "loss": 0.7903,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016818792930952328,
+      "loss": 0.8155,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016814231248463556,
+      "loss": 0.8403,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016809666917304603,
+      "loss": 0.8556,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016805099939249608,
+      "loss": 0.7637,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016800530316073747,
+      "loss": 0.8334,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016795958049553206,
+      "loss": 0.8682,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016791383141465222,
+      "loss": 0.7184,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016786805593588044,
+      "loss": 0.8616,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016782225407700952,
+      "loss": 0.7428,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016777642585584244,
+      "loss": 0.8472,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016773057129019253,
+      "loss": 0.7661,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016768469039788332,
+      "loss": 0.8225,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016763878319674854,
+      "loss": 0.7298,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016759284970463217,
+      "loss": 0.7384,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016754688993938843,
+      "loss": 0.7137,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016750090391888173,
+      "loss": 0.7234,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016745489166098673,
+      "loss": 0.7995,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016740885318358823,
+      "loss": 0.6672,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001673627885045812,
+      "loss": 0.7784,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016731669764187088,
+      "loss": 0.8314,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672705806133726,
+      "loss": 0.8116,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016722443743701196,
+      "loss": 0.6106,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671782681307246,
+      "loss": 0.7863,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016713207271245648,
+      "loss": 0.8368,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001670858512001635,
+      "loss": 0.7354,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001670396036118119,
+      "loss": 0.7581,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016699332996537793,
+      "loss": 0.9057,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000166947030278848,
+      "loss": 0.9363,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669007045702187,
+      "loss": 0.8121,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016685435285749667,
+      "loss": 0.7245,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016680797515869865,
+      "loss": 0.7541,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001667615714918515,
+      "loss": 0.7808,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016671514187499223,
+      "loss": 0.7609,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016666868632616785,
+      "loss": 0.7242,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016662220486343552,
+      "loss": 0.8291,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016657569750486238,
+      "loss": 0.8321,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001665291642685257,
+      "loss": 1.0469,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016648260517251285,
+      "loss": 0.7419,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016643602023492116,
+      "loss": 0.7175,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016638940947385806,
+      "loss": 0.7254,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016634277290744103,
+      "loss": 0.7041,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016629611055379756,
+      "loss": 0.7831,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016624942243106508,
+      "loss": 0.7112,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001662027085573912,
+      "loss": 0.8653,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016615596895093343,
+      "loss": 0.6677,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016610920362985932,
+      "loss": 0.7427,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016606241261234636,
+      "loss": 0.7167,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001660155959165821,
+      "loss": 0.8365,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016596875356076403,
+      "loss": 0.7868,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016592188556309968,
+      "loss": 0.6978,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016587499194180644,
+      "loss": 0.8189,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001658280727151117,
+      "loss": 0.7231,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016578112790125287,
+      "loss": 0.8298,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001657341575184772,
+      "loss": 0.8199,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016568716158504195,
+      "loss": 0.8182,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001656401401192143,
+      "loss": 0.7878,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001655930931392713,
+      "loss": 0.7336,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016554602066350002,
+      "loss": 0.801,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016549892271019737,
+      "loss": 0.9425,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016545179929767015,
+      "loss": 0.7378,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001654046504442351,
+      "loss": 0.8058,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016535747616821885,
+      "loss": 0.7045,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016531027648795786,
+      "loss": 0.6975,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016526305142179848,
+      "loss": 0.7823,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000165215800988097,
+      "loss": 0.8197,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016516852520521947,
+      "loss": 0.8039,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016512122409154185,
+      "loss": 0.8339,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016507389766544996,
+      "loss": 0.7519,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016502654594533937,
+      "loss": 0.7898,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016497916894961563,
+      "loss": 0.749,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016493176669669395,
+      "loss": 0.6982,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648843392049995,
+      "loss": 0.7349,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016483688649296713,
+      "loss": 0.7164,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016478940857904162,
+      "loss": 0.6656,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016474190548167744,
+      "loss": 0.8445,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016469437721933891,
+      "loss": 0.7495,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016464682381050016,
+      "loss": 0.7103,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016459924527364496,
+      "loss": 0.7961,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000164551641627267,
+      "loss": 0.7652,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001645040128898697,
+      "loss": 0.6445,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016445635907996612,
+      "loss": 0.7743,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016440868021607913,
+      "loss": 0.696,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001643609763167415,
+      "loss": 0.6986,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016431324740049538,
+      "loss": 0.7327,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016426549348589302,
+      "loss": 0.9221,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016421771459149614,
+      "loss": 0.7454,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641699107358763,
+      "loss": 0.7889,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641220819376146,
+      "loss": 0.8234,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016407422821530204,
+      "loss": 0.6964,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640263495875392,
+      "loss": 0.6772,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016397844607293627,
+      "loss": 0.7424,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001639305176901133,
+      "loss": 0.7814,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016388256445769986,
+      "loss": 0.7978,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016383458639433523,
+      "loss": 0.8468,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637865835186683,
+      "loss": 0.8618,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016373855584935764,
+      "loss": 0.7545,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001636905034050715,
+      "loss": 0.7417,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016364242620448767,
+      "loss": 0.7274,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016359432426629362,
+      "loss": 0.7907,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635461976091864,
+      "loss": 0.7878,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016349804625187276,
+      "loss": 0.757,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634498702130689,
+      "loss": 0.7382,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634016695115007,
+      "loss": 0.7082,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001633534441659037,
+      "loss": 0.7155,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001633051941950228,
+      "loss": 0.7722,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001632569196176128,
+      "loss": 0.8197,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016320862045243772,
+      "loss": 0.7446,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016316029671827135,
+      "loss": 0.7369,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016311194843389697,
+      "loss": 0.6842,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630635756181074,
+      "loss": 0.8074,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016301517828970502,
+      "loss": 0.7622,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001629667564675017,
+      "loss": 0.9159,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016291831017031883,
+      "loss": 0.71,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016286983941698736,
+      "loss": 0.6926,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016282134422634776,
+      "loss": 0.6807,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016277282461724988,
+      "loss": 0.7957,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016272428060855315,
+      "loss": 0.8029,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001626757122191265,
+      "loss": 0.7787,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016262711946784833,
+      "loss": 0.7094,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016257850237360647,
+      "loss": 0.8329,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016252986095529817,
+      "loss": 0.7218,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016248119523183031,
+      "loss": 0.7285,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016243250522211903,
+      "loss": 0.7857,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016238379094509,
+      "loss": 0.7518,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016233505241967827,
+      "loss": 0.7893,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016228628966482843,
+      "loss": 0.7574,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016223750269949433,
+      "loss": 0.6416,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016218869154263933,
+      "loss": 0.816,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016213985621323622,
+      "loss": 0.8039,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016209099673026708,
+      "loss": 0.7018,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016204211311272347,
+      "loss": 0.5592,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016199320537960629,
+      "loss": 0.7157,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619442735499258,
+      "loss": 0.7487,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016189531764270167,
+      "loss": 0.7669,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618463376769629,
+      "loss": 0.6992,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617973336717479,
+      "loss": 0.7649,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016174830564610426,
+      "loss": 0.8041,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016169925361908912,
+      "loss": 0.7722,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616501776097688,
+      "loss": 0.6872,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000161601077637219,
+      "loss": 0.8237,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001615519537205247,
+      "loss": 0.7161,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001615028058787803,
+      "loss": 0.8523,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016145363413108926,
+      "loss": 0.7735,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016140443849656464,
+      "loss": 0.7142,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016135521899432855,
+      "loss": 0.8298,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016130597564351245,
+      "loss": 0.8147,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016125670846325706,
+      "loss": 0.6923,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016120741747271248,
+      "loss": 0.7739,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001611581026910378,
+      "loss": 0.719,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016110876413740166,
+      "loss": 0.7466,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016105940183098176,
+      "loss": 0.7608,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016101001579096503,
+      "loss": 0.7907,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016096060603654772,
+      "loss": 0.7067,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001609111725869352,
+      "loss": 0.6229,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016086171546134215,
+      "loss": 0.7406,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016081223467899236,
+      "loss": 0.8798,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016076273025911886,
+      "loss": 0.8117,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016071320222096387,
+      "loss": 0.7549,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016066365058377876,
+      "loss": 0.7478,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016061407536682411,
+      "loss": 0.7596,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016056447658936967,
+      "loss": 0.655,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016051485427069433,
+      "loss": 0.7824,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604652084300861,
+      "loss": 0.6582,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016041553908684213,
+      "loss": 0.8531,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001603658462602688,
+      "loss": 0.8498,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016031612996968156,
+      "loss": 0.8003,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602663902344049,
+      "loss": 0.7997,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016021662707377257,
+      "loss": 0.7522,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016016684050712736,
+      "loss": 0.7143,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016011703055382107,
+      "loss": 0.7986,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016006719723321474,
+      "loss": 0.7892,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016001734056467844,
+      "loss": 0.7703,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015996746056759123,
+      "loss": 0.6789,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015991755726134135,
+      "loss": 0.7313,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015986763066532607,
+      "loss": 0.6811,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015981768079895167,
+      "loss": 0.8308,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597677076816335,
+      "loss": 0.8018,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000159717711332796,
+      "loss": 0.7975,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015966769177187256,
+      "loss": 0.7538,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015961764901830562,
+      "loss": 0.7313,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015956758309154662,
+      "loss": 0.7785,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595174940110561,
+      "loss": 0.7644,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001594673817963035,
+      "loss": 0.6795,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015941724646676724,
+      "loss": 0.7244,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015936708804193477,
+      "loss": 0.8051,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015931690654130256,
+      "loss": 0.7681,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015926670198437596,
+      "loss": 0.8116,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015921647439066933,
+      "loss": 0.7311,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015916622377970597,
+      "loss": 0.7027,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015911595017101816,
+      "loss": 0.7291,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590656535841471,
+      "loss": 0.9054,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015901533403864286,
+      "loss": 0.7526,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015896499155406451,
+      "loss": 0.8606,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589146261499801,
+      "loss": 0.7651,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015886423784596635,
+      "loss": 0.7812,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015881382666160915,
+      "loss": 0.7139,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015876339261650314,
+      "loss": 0.8502,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015871293573025186,
+      "loss": 0.784,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015866245602246773,
+      "loss": 0.7915,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586119535127721,
+      "loss": 0.8762,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585614282207951,
+      "loss": 0.7778,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015851088016617575,
+      "loss": 0.7513,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015846030936856193,
+      "loss": 0.6368,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015840971584761036,
+      "loss": 0.7806,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015835909962298654,
+      "loss": 0.6628,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015830846071436492,
+      "loss": 0.8351,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001582577991414286,
+      "loss": 0.7638,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015820711492386963,
+      "loss": 0.6775,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015815640808138873,
+      "loss": 0.688,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001581056786336956,
+      "loss": 0.7867,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580549266005085,
+      "loss": 0.8594,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580041520015547,
+      "loss": 0.7284,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015795335485657,
+      "loss": 0.7152,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015790253518529923,
+      "loss": 0.7527,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015785169300749574,
+      "loss": 0.7192,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015780082834292172,
+      "loss": 0.7394,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015774994121134817,
+      "loss": 0.6151,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015769903163255467,
+      "loss": 0.8639,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015764809962632968,
+      "loss": 0.6641,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015759714521247035,
+      "loss": 0.8125,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001575461684107824,
+      "loss": 0.7513,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574951692410804,
+      "loss": 0.7436,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015744414772318754,
+      "loss": 0.6203,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015739310387693576,
+      "loss": 0.8351,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015734203772216563,
+      "loss": 0.7663,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001572909492787264,
+      "loss": 0.6689,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015723983856647596,
+      "loss": 0.7381,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015718870560528094,
+      "loss": 0.7633,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015713755041501653,
+      "loss": 0.8251,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015708637301556656,
+      "loss": 0.823,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015703517342682358,
+      "loss": 0.801,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015698395166868868,
+      "loss": 0.6433,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015693270776107157,
+      "loss": 0.6339,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015688144172389066,
+      "loss": 0.6623,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015683015357707283,
+      "loss": 0.6422,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567788433405537,
+      "loss": 0.7793,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015672751103427732,
+      "loss": 0.7017,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001566761566781964,
+      "loss": 0.7893,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015662478029227227,
+      "loss": 0.7508,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565733818964747,
+      "loss": 0.7784,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015652196151078217,
+      "loss": 0.803,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015647051915518154,
+      "loss": 0.7341,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015641905484966833,
+      "loss": 0.7286,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015636756861424656,
+      "loss": 0.7073,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015631606046892876,
+      "loss": 0.7416,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015626453043373598,
+      "loss": 0.6171,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015621297852869777,
+      "loss": 0.7733,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561614047738522,
+      "loss": 0.7981,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015610980918924586,
+      "loss": 0.8156,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015605819179493374,
+      "loss": 0.6444,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015600655261097937,
+      "loss": 0.806,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015595489165745474,
+      "loss": 0.7601,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015590320895444033,
+      "loss": 0.8205,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015585150452202502,
+      "loss": 0.7548,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015579977838030615,
+      "loss": 0.927,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001557480305493895,
+      "loss": 0.7469,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015569626104938933,
+      "loss": 0.8441,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001556444699004283,
+      "loss": 0.7751,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015559265712263736,
+      "loss": 0.73,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015554082273615608,
+      "loss": 0.7726,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015548896676113223,
+      "loss": 0.6648,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015543708921772215,
+      "loss": 0.7584,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015538519012609043,
+      "loss": 0.7205,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001553332695064101,
+      "loss": 0.6482,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015528132737886252,
+      "loss": 0.7729,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015522936376363746,
+      "loss": 0.7127,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015517737868093301,
+      "loss": 0.7848,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015512537215095558,
+      "loss": 0.7208,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015507334419391996,
+      "loss": 0.6785,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015502129483004927,
+      "loss": 0.6806,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549692240795749,
+      "loss": 0.758,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549171319627366,
+      "loss": 0.7869,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015486501849978242,
+      "loss": 0.6598,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548128837109687,
+      "loss": 0.8078,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001547607276165601,
+      "loss": 0.7352,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015470855023682942,
+      "loss": 0.7801,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015465635159205796,
+      "loss": 0.7019,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015460413170253508,
+      "loss": 0.8488,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015455189058855858,
+      "loss": 0.803,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015449962827043435,
+      "loss": 0.7498,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015444734476847657,
+      "loss": 0.7097,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015439504010300774,
+      "loss": 0.7475,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001543427142943585,
+      "loss": 0.6231,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015429036736286765,
+      "loss": 0.7602,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001542379993288824,
+      "loss": 0.7632,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015418561021275792,
+      "loss": 0.8102,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015413320003485777,
+      "loss": 0.9521,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015408076881555363,
+      "loss": 0.6995,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001540283165752253,
+      "loss": 0.8195,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001539758433342608,
+      "loss": 0.8016,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001539233491130564,
+      "loss": 0.8106,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001538708339320163,
+      "loss": 0.7386,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015381829781155307,
+      "loss": 0.738,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015376574077208735,
+      "loss": 0.7724,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015371316283404785,
+      "loss": 0.7019,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001536605640178715,
+      "loss": 0.6876,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015360794434400322,
+      "loss": 0.8201,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015355530383289617,
+      "loss": 0.6638,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015350264250501154,
+      "loss": 0.7319,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001534499603808186,
+      "loss": 0.7264,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533972574807947,
+      "loss": 0.7906,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533445338254254,
+      "loss": 0.7186,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001532917894352041,
+      "loss": 0.7205,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015323902433063245,
+      "loss": 0.6238,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015318623853222005,
+      "loss": 0.6979,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015313343206048457,
+      "loss": 0.8194,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015308060493595172,
+      "loss": 0.7184,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015302775717915528,
+      "loss": 0.8205,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015297488881063696,
+      "loss": 0.7036,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015292199985094654,
+      "loss": 0.7571,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015286909032064173,
+      "loss": 0.7272,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015281616024028844,
+      "loss": 0.6205,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001527632096304603,
+      "loss": 0.6794,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015271023851173908,
+      "loss": 0.7151,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015265724690471455,
+      "loss": 0.8552,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015260423482998427,
+      "loss": 0.7302,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015255120230815394,
+      "loss": 0.7699,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015249814935983712,
+      "loss": 0.7743,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015244507600565532,
+      "loss": 0.6218,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015239198226623798,
+      "loss": 0.7764,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015233886816222247,
+      "loss": 0.7175,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001522857337142541,
+      "loss": 0.9022,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000152232578942986,
+      "loss": 0.7764,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015217940386907933,
+      "loss": 0.7706,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015212620851320306,
+      "loss": 0.7909,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000152072992896034,
+      "loss": 0.6567,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015201975703825696,
+      "loss": 0.8343,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519665009605645,
+      "loss": 0.6977,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015191322468365717,
+      "loss": 0.6601,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015185992822824317,
+      "loss": 0.7231,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015180661161503877,
+      "loss": 0.772,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001517532748647679,
+      "loss": 0.711,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015169991799816244,
+      "loss": 0.8243,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015164654103596204,
+      "loss": 0.7524,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015159314399891405,
+      "loss": 0.7547,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015153972690777384,
+      "loss": 0.8354,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015148628978330444,
+      "loss": 0.7192,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514328326462767,
+      "loss": 0.7465,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001513793555174692,
+      "loss": 1.0077,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015132585841766836,
+      "loss": 0.959,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512723413676683,
+      "loss": 0.7721,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000151218804388271,
+      "loss": 0.7243,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015116524750028606,
+      "loss": 0.8358,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001511116707245309,
+      "loss": 0.7296,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001510580740818306,
+      "loss": 0.7766,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015100445759301805,
+      "loss": 0.8524,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015095082127893376,
+      "loss": 0.7684,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508971651604261,
+      "loss": 0.74,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508434892583509,
+      "loss": 0.6579,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015078979359357193,
+      "loss": 0.8238,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015073607818696043,
+      "loss": 0.7815,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015068234305939547,
+      "loss": 0.8297,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015062858823176374,
+      "loss": 0.7246,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015057481372495948,
+      "loss": 0.8754,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015052101955988478,
+      "loss": 0.7063,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015046720575744916,
+      "loss": 0.8742,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015041337233856998,
+      "loss": 0.8392,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015035951932417205,
+      "loss": 0.9192,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001503056467351879,
+      "loss": 0.6879,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001502517545925576,
+      "loss": 0.7009,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015019784291722888,
+      "loss": 0.7481,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015014391173015702,
+      "loss": 0.7566,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015008996105230494,
+      "loss": 0.8241,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015003599090464306,
+      "loss": 0.6804,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014998200130814942,
+      "loss": 0.6755,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001499279922838096,
+      "loss": 0.7344,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014987396385261677,
+      "loss": 0.7353,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014981991603557154,
+      "loss": 0.6317,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014976584885368222,
+      "loss": 0.7148,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014971176232796447,
+      "loss": 0.5827,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014965765647944162,
+      "loss": 0.7642,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014960353132914436,
+      "loss": 0.7284,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014954938689811106,
+      "loss": 0.6644,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494952232073874,
+      "loss": 0.7119,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014944104027802668,
+      "loss": 0.8273,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014938683813108967,
+      "loss": 0.6533,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001493326167876445,
+      "loss": 0.8051,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014927837626876688,
+      "loss": 0.6984,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492241165955399,
+      "loss": 0.7123,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001491698377890542,
+      "loss": 0.6347,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014911553987040767,
+      "loss": 0.7001,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001490612228607058,
+      "loss": 0.6899,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014900688678106145,
+      "loss": 0.6505,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014895253165259485,
+      "loss": 0.7253,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488981574964337,
+      "loss": 0.6782,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014884376433371303,
+      "loss": 0.7013,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487893521855753,
+      "loss": 0.7741,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014873492107317037,
+      "loss": 0.8162,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001486804710176554,
+      "loss": 0.6369,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014862600204019495,
+      "loss": 0.6969,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148571514161961,
+      "loss": 0.861,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014851700740413278,
+      "loss": 0.7567,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001484624817878969,
+      "loss": 0.7467,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014840793733444726,
+      "loss": 0.7591,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014835337406498514,
+      "loss": 0.7752,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014829879200071915,
+      "loss": 0.7767,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014824419116286515,
+      "loss": 0.6721,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014818957157264624,
+      "loss": 0.8212,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014813493325129294,
+      "loss": 0.8803,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148080276220043,
+      "loss": 0.7382,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014802560050014145,
+      "loss": 0.8429,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014797090611284053,
+      "loss": 0.8313,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001479161930793998,
+      "loss": 0.7541,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014786146142108602,
+      "loss": 0.7091,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014780671115917324,
+      "loss": 0.8195,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014775194231494268,
+      "loss": 0.7467,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014769715490968287,
+      "loss": 0.7329,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014764234896468944,
+      "loss": 0.6953,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014758752450126536,
+      "loss": 0.783,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014753268154072067,
+      "loss": 0.6513,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014747782010437262,
+      "loss": 0.7084,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014742294021354577,
+      "loss": 0.7605,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014736804188957174,
+      "loss": 0.7505,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001473131251537893,
+      "loss": 0.7181,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001472581900275444,
+      "loss": 0.6259,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014720323653219018,
+      "loss": 0.8511,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001471482646890869,
+      "loss": 0.7474,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001470932745196019,
+      "loss": 0.6903,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014703826604510977,
+      "loss": 0.7874,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014698323928699202,
+      "loss": 0.6802,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014692819426663744,
+      "loss": 0.7089,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014687313100544187,
+      "loss": 0.7343,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014681804952480817,
+      "loss": 0.8292,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467629498461464,
+      "loss": 0.7768,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467078319908736,
+      "loss": 0.8526,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001466526959804139,
+      "loss": 0.7601,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014659754183619846,
+      "loss": 0.8336,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014654236957966558,
+      "loss": 0.7828,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014648717923226057,
+      "loss": 0.7006,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014643197081543564,
+      "loss": 0.7285,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014637674435065022,
+      "loss": 0.7699,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014632149985937062,
+      "loss": 0.8399,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014626623736307018,
+      "loss": 0.8453,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014621095688322933,
+      "loss": 0.7084,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001461556584413354,
+      "loss": 0.7551,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014610034205888263,
+      "loss": 0.6944,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001460450077573724,
+      "loss": 0.7258,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014598965555831297,
+      "loss": 0.6778,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014593428548321957,
+      "loss": 0.7389,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014587889755361436,
+      "loss": 0.7671,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014582349179102644,
+      "loss": 0.6161,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014576806821699187,
+      "loss": 0.8267,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014571262685305364,
+      "loss": 0.744,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014565716772076161,
+      "loss": 0.759,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014560169084167258,
+      "loss": 0.7713,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014554619623735021,
+      "loss": 0.866,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014549068392936515,
+      "loss": 0.7223,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014543515393929473,
+      "loss": 0.7853,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001453796062887234,
+      "loss": 0.9289,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001453240409992423,
+      "loss": 0.8365,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014526845809244952,
+      "loss": 0.7293,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014521285758994993,
+      "loss": 0.7119,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014515723951335528,
+      "loss": 0.6778,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014510160388428412,
+      "loss": 0.6352,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001450459507243619,
+      "loss": 0.733,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001449902800552208,
+      "loss": 0.7194,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014493459189849977,
+      "loss": 0.696,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014487888627584468,
+      "loss": 0.8067,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014482316320890815,
+      "loss": 0.785,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014476742271934953,
+      "loss": 0.7678,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000144711664828835,
+      "loss": 0.8259,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014465588955903745,
+      "loss": 0.8334,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446000969316365,
+      "loss": 0.6748,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014454428696831869,
+      "loss": 0.627,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014448845969077705,
+      "loss": 0.7804,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014443261512071156,
+      "loss": 0.68,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001443767532798288,
+      "loss": 0.7341,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014432087418984207,
+      "loss": 0.8002,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001442649778724714,
+      "loss": 0.806,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014420906434944354,
+      "loss": 0.6571,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014415313364249188,
+      "loss": 0.6462,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014409718577335654,
+      "loss": 0.8077,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014404122076378423,
+      "loss": 0.6381,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439852386355284,
+      "loss": 0.6846,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439292394103491,
+      "loss": 0.7696,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014387322311001309,
+      "loss": 0.7535,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014381718975629374,
+      "loss": 0.7853,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000143761139370971,
+      "loss": 0.7534,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014370507197583146,
+      "loss": 0.7276,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014364898759266836,
+      "loss": 0.7422,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014359288624328156,
+      "loss": 0.7885,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014353676794947746,
+      "loss": 0.8337,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014348063273306904,
+      "loss": 0.7594,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001434244806158759,
+      "loss": 0.7273,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014336831161972417,
+      "loss": 0.7201,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014331212576644656,
+      "loss": 0.8107,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014325592307788237,
+      "loss": 0.7039,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014319970357587743,
+      "loss": 0.8587,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014314346728228397,
+      "loss": 0.6796,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014308721421896098,
+      "loss": 0.7111,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001430309444077738,
+      "loss": 0.6371,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014297465787059427,
+      "loss": 0.7689,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001429183546293009,
+      "loss": 0.7617,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014286203470577854,
+      "loss": 0.8539,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001428056981219185,
+      "loss": 0.6761,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001427493448996187,
+      "loss": 0.6937,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014269297506078348,
+      "loss": 0.7384,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014263658862732358,
+      "loss": 0.7609,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014258018562115623,
+      "loss": 0.7616,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001425237660642051,
+      "loss": 0.7372,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014246732997840028,
+      "loss": 0.8174,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014241087738567837,
+      "loss": 0.673,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014235440830798218,
+      "loss": 0.7205,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014229792276726123,
+      "loss": 0.7841,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014224142078547112,
+      "loss": 0.7071,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014218490238457408,
+      "loss": 0.7965,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014212836758653862,
+      "loss": 0.7732,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014207181641333962,
+      "loss": 0.6746,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014201524888695835,
+      "loss": 0.6607,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014195866502938245,
+      "loss": 0.7496,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014190206486260583,
+      "loss": 0.8088,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014184544840862887,
+      "loss": 0.8118,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014178881568945815,
+      "loss": 0.6118,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014173216672710664,
+      "loss": 0.7882,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014167550154359364,
+      "loss": 0.6688,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014161882016094473,
+      "loss": 0.76,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014156212260119176,
+      "loss": 0.6985,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014150540888637286,
+      "loss": 0.7135,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014144867903853253,
+      "loss": 0.7565,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014139193307972147,
+      "loss": 0.7738,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014133517103199664,
+      "loss": 0.5833,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014127839291742128,
+      "loss": 0.6498,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014122159875806488,
+      "loss": 0.784,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001411647885760031,
+      "loss": 0.6274,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014110796239331791,
+      "loss": 0.7672,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001410511202320975,
+      "loss": 0.749,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001409942621144362,
+      "loss": 0.5874,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014093738806243456,
+      "loss": 0.7799,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001408804980981994,
+      "loss": 0.6992,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001408235922438436,
+      "loss": 0.807,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014076667052148636,
+      "loss": 0.7891,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014070973295325293,
+      "loss": 0.7456,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014065277956127478,
+      "loss": 0.7583,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001405958103676895,
+      "loss": 0.6642,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014053882539464082,
+      "loss": 0.9421,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001404818246642787,
+      "loss": 0.7015,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014042480819875908,
+      "loss": 0.8836,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014036777602024404,
+      "loss": 0.7865,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014031072815090187,
+      "loss": 0.8391,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001402536646129069,
+      "loss": 0.7448,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014019658542843951,
+      "loss": 0.7331,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014013949061968624,
+      "loss": 0.7467,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014008238020883962,
+      "loss": 0.6634,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014002525421809832,
+      "loss": 0.7337,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013996811266966702,
+      "loss": 0.6452,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013991095558575645,
+      "loss": 0.7158,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001398537829885834,
+      "loss": 0.7217,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013979659490037064,
+      "loss": 0.7231,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013973939134334708,
+      "loss": 0.7432,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001396821723397475,
+      "loss": 0.7255,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013962493791181275,
+      "loss": 0.764,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001395676880817897,
+      "loss": 0.6336,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013951042287193118,
+      "loss": 0.7468,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013945314230449593,
+      "loss": 0.6217,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001393958464017488,
+      "loss": 0.7797,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013933853518596046,
+      "loss": 0.7679,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001392812086794077,
+      "loss": 0.7928,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013922386690437305,
+      "loss": 0.7696,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013916650988314515,
+      "loss": 0.7009,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013910913763801844,
+      "loss": 0.7056,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013905175019129337,
+      "loss": 0.7154,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013899434756527623,
+      "loss": 0.7789,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013893692978227923,
+      "loss": 0.7467,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013887949686462053,
+      "loss": 0.6573,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001388220488346241,
+      "loss": 0.7845,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013876458571461978,
+      "loss": 0.7906,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013870710752694333,
+      "loss": 0.8264,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013864961429393632,
+      "loss": 0.6762,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013859210603794626,
+      "loss": 0.9264,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013853458278132637,
+      "loss": 0.7815,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001384770445464357,
+      "loss": 0.6656,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013841949135563929,
+      "loss": 0.8143,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013836192323130782,
+      "loss": 0.8267,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013830434019581787,
+      "loss": 0.7526,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013824674227155175,
+      "loss": 0.7701,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381891294808976,
+      "loss": 0.6944,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013813150184624934,
+      "loss": 0.7191,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013807385939000667,
+      "loss": 0.7324,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000138016202134575,
+      "loss": 0.8004,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013795853010236555,
+      "loss": 0.7464,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001379008433157952,
+      "loss": 0.6781,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013784314179728668,
+      "loss": 0.7423,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013778542556926838,
+      "loss": 0.8042,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013772769465417438,
+      "loss": 0.7192,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001376699490744446,
+      "loss": 0.8419,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013761218885252442,
+      "loss": 0.6456,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001375544140108652,
+      "loss": 0.7073,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001374966245719237,
+      "loss": 0.7337,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013743882055816258,
+      "loss": 0.7786,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013738100199205006,
+      "loss": 0.8048,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013732316889606005,
+      "loss": 0.6198,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013726532129267207,
+      "loss": 0.7351,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013720745920437128,
+      "loss": 0.8299,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013714958265364854,
+      "loss": 0.7437,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013709169166300028,
+      "loss": 0.8636,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013703378625492852,
+      "loss": 0.7544,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013697586645194088,
+      "loss": 0.7443,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013691793227655065,
+      "loss": 0.7546,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001368599837512767,
+      "loss": 0.8752,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013680202089864333,
+      "loss": 0.7841,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001367440437411806,
+      "loss": 0.6778,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013668605230142403,
+      "loss": 0.774,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013662804660191468,
+      "loss": 0.7416,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013657002666519924,
+      "loss": 0.6704,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013651199251382983,
+      "loss": 0.6715,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013645394417036416,
+      "loss": 0.6724,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013639588165736543,
+      "loss": 0.7623,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013633780499740236,
+      "loss": 0.8233,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013627971421304914,
+      "loss": 0.7681,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013622160932688553,
+      "loss": 0.7837,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013616349036149668,
+      "loss": 0.7945,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013610535733947325,
+      "loss": 0.7784,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013604721028341135,
+      "loss": 0.698,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001359890492159126,
+      "loss": 0.7327,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013593087415958397,
+      "loss": 0.8148,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013587268513703794,
+      "loss": 0.7282,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013581448217089243,
+      "loss": 0.7438,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013575626528377073,
+      "loss": 0.7693,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013569803449830153,
+      "loss": 0.7253,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013563978983711897,
+      "loss": 0.676,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001355815313228626,
+      "loss": 0.6757,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013552325897817728,
+      "loss": 0.6961,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001354649728257133,
+      "loss": 0.7355,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013540667288812632,
+      "loss": 0.7776,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013534835918807732,
+      "loss": 0.7775,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013529003174823264,
+      "loss": 0.8042,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000135231690591264,
+      "loss": 0.844,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001351733357398484,
+      "loss": 0.7503,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013511496721666823,
+      "loss": 0.8078,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001350565850444111,
+      "loss": 0.7597,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013499818924576999,
+      "loss": 0.786,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013493977984344318,
+      "loss": 0.7803,
+      "step": 2103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001348813568601342,
+      "loss": 0.77,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001348229203185519,
+      "loss": 0.7512,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013476447024141037,
+      "loss": 0.7063,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013470600665142893,
+      "loss": 0.798,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013464752957133226,
+      "loss": 0.7717,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013458903902385019,
+      "loss": 0.6673,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013453053503171782,
+      "loss": 0.8273,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013447201761767542,
+      "loss": 0.6526,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001344134868044686,
+      "loss": 0.6792,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013435494261484804,
+      "loss": 0.7592,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013429638507156969,
+      "loss": 0.7273,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013423781419739472,
+      "loss": 0.7973,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013417923001508942,
+      "loss": 0.7167,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013412063254742527,
+      "loss": 0.7543,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001340620218171789,
+      "loss": 0.7278,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001340033978471322,
+      "loss": 0.7232,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013394476066007203,
+      "loss": 0.7353,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013388611027879052,
+      "loss": 0.8289,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013382744672608488,
+      "loss": 0.6483,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013376877002475746,
+      "loss": 0.7791,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013371008019761568,
+      "loss": 0.8034,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013365137726747213,
+      "loss": 0.8043,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013359266125714442,
+      "loss": 0.736,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013353393218945532,
+      "loss": 0.7438,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013347519008723258,
+      "loss": 0.7265,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001334164349733091,
+      "loss": 0.6621,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001333576668705228,
+      "loss": 0.686,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013329888580171666,
+      "loss": 0.63,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001332400917897387,
+      "loss": 0.8008,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013318128485744198,
+      "loss": 0.8005,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013312246502768452,
+      "loss": 0.6978,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013306363232332944,
+      "loss": 0.7225,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013300478676724482,
+      "loss": 0.7212,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013294592838230374,
+      "loss": 0.7183,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013288705719138427,
+      "loss": 0.7387,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013282817321736947,
+      "loss": 0.7521,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013276927648314733,
+      "loss": 0.721,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013271036701161082,
+      "loss": 0.6997,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001326514448256579,
+      "loss": 0.7466,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001325925099481914,
+      "loss": 0.7878,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013253356240211915,
+      "loss": 0.7681,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013247460221035387,
+      "loss": 0.8955,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001324156293958132,
+      "loss": 0.7759,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013235664398141968,
+      "loss": 0.6927,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013229764599010076,
+      "loss": 0.9109,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013223863544478874,
+      "loss": 0.7555,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321796123684209,
+      "loss": 0.6075,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013212057678393924,
+      "loss": 0.6559,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001320615287142908,
+      "loss": 0.732,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013200246818242733,
+      "loss": 0.7085,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013194339521130542,
+      "loss": 0.7163,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013188430982388665,
+      "loss": 0.7002,
+      "step": 2155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013182521204313725,
+      "loss": 0.666,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013176610189202836,
+      "loss": 0.6214,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013170697939353593,
+      "loss": 0.8016,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013164784457064065,
+      "loss": 0.7779,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013158869744632807,
+      "loss": 0.6366,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001315295380435885,
+      "loss": 0.6731,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013147036638541697,
+      "loss": 0.7675,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013141118249481335,
+      "loss": 0.6697,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013135198639478222,
+      "loss": 0.6897,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013129277810833292,
+      "loss": 0.747,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001312335576584795,
+      "loss": 0.7579,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013117432506824084,
+      "loss": 0.6512,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013111508036064038,
+      "loss": 0.6808,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013105582355870635,
+      "loss": 0.7395,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013099655468547174,
+      "loss": 0.7706,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013093727376397413,
+      "loss": 0.677,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013087798081725586,
+      "loss": 0.7588,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013081867586836387,
+      "loss": 0.6238,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013075935894034984,
+      "loss": 0.7531,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013070003005627006,
+      "loss": 0.7241,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001306406892391855,
+      "loss": 0.7217,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001305813365121617,
+      "loss": 0.6994,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001305219718982689,
+      "loss": 0.7786,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013046259542058198,
+      "loss": 0.6589,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013040320710218034,
+      "loss": 0.734,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013034380696614806,
+      "loss": 0.7175,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013028439503557374,
+      "loss": 0.7573,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013022497133355064,
+      "loss": 0.7235,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001301655358831766,
+      "loss": 0.7823,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013010608870755395,
+      "loss": 0.7301,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013004662982978962,
+      "loss": 0.7108,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012998715927299513,
+      "loss": 0.6768,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012992767706028642,
+      "loss": 0.8074,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001298681832147841,
+      "loss": 0.7518,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012980867775961322,
+      "loss": 0.7241,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012974916071790336,
+      "loss": 0.7287,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012968963211278865,
+      "loss": 0.6418,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001296300919674076,
+      "loss": 0.8016,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012957054030490332,
+      "loss": 0.7258,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012951097714842332,
+      "loss": 0.782,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012945140252111967,
+      "loss": 0.7642,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001293918164461488,
+      "loss": 0.7723,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001293322189466716,
+      "loss": 0.7218,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012927261004585348,
+      "loss": 0.666,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001292129897668642,
+      "loss": 0.8446,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012915335813287802,
+      "loss": 0.6781,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012909371516707353,
+      "loss": 0.7122,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012903406089263375,
+      "loss": 0.7443,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012897439533274613,
+      "loss": 0.7807,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012891471851060252,
+      "loss": 0.6796,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000128855030449399,
+      "loss": 0.7615,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001287953311723363,
+      "loss": 0.7064,
+      "step": 2207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012873562070261918,
+      "loss": 0.71,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012867589906345698,
+      "loss": 0.8468,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012861616627806336,
+      "loss": 0.7134,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012855642236965614,
+      "loss": 0.7348,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012849666736145773,
+      "loss": 0.7495,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001284369012766946,
+      "loss": 0.7221,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012837712413859772,
+      "loss": 0.7097,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012831733597040224,
+      "loss": 0.7628,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012825753679534762,
+      "loss": 0.6911,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012819772663667762,
+      "loss": 0.7401,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001281379055176403,
+      "loss": 0.7281,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012807807346148784,
+      "loss": 0.7957,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012801823049147688,
+      "loss": 0.7088,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001279583766308681,
+      "loss": 0.7051,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012789851190292664,
+      "loss": 0.8523,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001278386363309216,
+      "loss": 0.7732,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012777874993812646,
+      "loss": 0.8871,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001277188527478189,
+      "loss": 0.5935,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012765894478328072,
+      "loss": 0.6552,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012759902606779804,
+      "loss": 0.7425,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000127539096624661,
+      "loss": 0.7582,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012747915647716402,
+      "loss": 0.7477,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001274192056486056,
+      "loss": 0.8559,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012735924416228852,
+      "loss": 0.6409,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012729927204151956,
+      "loss": 0.6092,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001272392893096097,
+      "loss": 0.79,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012717929598987405,
+      "loss": 0.7457,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012711929210563187,
+      "loss": 0.768,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001270592776802064,
+      "loss": 0.7928,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012699925273692507,
+      "loss": 0.7092,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012693921729911938,
+      "loss": 0.7991,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000126879171390125,
+      "loss": 0.6725,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012681911503328153,
+      "loss": 0.7405,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012675904825193264,
+      "loss": 0.784,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012669897106942614,
+      "loss": 0.7673,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001266388835091139,
+      "loss": 0.6558,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012657878559435166,
+      "loss": 0.7482,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012651867734849934,
+      "loss": 0.6982,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012645855879492085,
+      "loss": 0.7369,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000126398429956984,
+      "loss": 0.8799,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001263382908580608,
+      "loss": 0.6032,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012627814152152701,
+      "loss": 0.7926,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012621798197076257,
+      "loss": 0.6484,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001261578122291513,
+      "loss": 0.6412,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001260976323200809,
+      "loss": 0.7524,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012603744226694324,
+      "loss": 0.6768,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012597724209313391,
+      "loss": 0.7503,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012591703182205256,
+      "loss": 0.6823,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012585681147710275,
+      "loss": 0.7198,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001257965810816919,
+      "loss": 0.7839,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001257363406592314,
+      "loss": 0.7533,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012567609023313647,
+      "loss": 0.6778,
+      "step": 2259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012561582982682635,
+      "loss": 0.58,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000125555559463724,
+      "loss": 0.8624,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012549527916725633,
+      "loss": 0.7516,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012543498896085408,
+      "loss": 0.7002,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001253746888679519,
+      "loss": 0.7684,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012531437891198825,
+      "loss": 0.7748,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012525405911640542,
+      "loss": 0.646,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012519372950464948,
+      "loss": 0.8106,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001251333901001704,
+      "loss": 0.6033,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001250730409264219,
+      "loss": 0.8774,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012501268200686152,
+      "loss": 0.7361,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012495231336495058,
+      "loss": 0.6529,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001248919350241542,
+      "loss": 0.7508,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012483154700794122,
+      "loss": 0.7854,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012477114933978427,
+      "loss": 0.7277,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012471074204315976,
+      "loss": 0.7044,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012465032514154777,
+      "loss": 0.733,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012458989865843224,
+      "loss": 0.8063,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012452946261730068,
+      "loss": 0.7079,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001244690170416444,
+      "loss": 0.7408,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001244085619549584,
+      "loss": 0.833,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012434809738074142,
+      "loss": 0.7811,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012428762334249584,
+      "loss": 0.6451,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012422713986372772,
+      "loss": 0.7351,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012416664696794676,
+      "loss": 0.7639,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001241061446786664,
+      "loss": 0.7294,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012404563301940367,
+      "loss": 0.6678,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012398511201367927,
+      "loss": 0.7566,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012392458168501754,
+      "loss": 0.7784,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001238640420569464,
+      "loss": 0.804,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012380349315299744,
+      "loss": 0.7838,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001237429349967058,
+      "loss": 0.8297,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012368236761161022,
+      "loss": 0.6665,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012362179102125317,
+      "loss": 0.6522,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012356120524918045,
+      "loss": 0.8559,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001235006103189416,
+      "loss": 0.6939,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012344000625408964,
+      "loss": 0.763,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012337939307818126,
+      "loss": 0.6762,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012331877081477662,
+      "loss": 0.666,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001232581394874393,
+      "loss": 0.7549,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012319749911973658,
+      "loss": 0.8901,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012313684973523913,
+      "loss": 0.7643,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012307619135752123,
+      "loss": 0.7005,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012301552401016057,
+      "loss": 0.7278,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012295484771673837,
+      "loss": 0.6954,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001228941625008393,
+      "loss": 0.7567,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012283346838605147,
+      "loss": 0.8114,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001227727653959666,
+      "loss": 0.7484,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012271205355417965,
+      "loss": 0.7398,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001226513328842892,
+      "loss": 0.7667,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012259060340989708,
+      "loss": 0.5695,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012252986515460878,
+      "loss": 0.7962,
+      "step": 2311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012246911814203294,
+      "loss": 0.6375,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001224083623957818,
+      "loss": 0.6868,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001223475979394709,
+      "loss": 0.8868,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012228682479671927,
+      "loss": 0.6763,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001222260429911491,
+      "loss": 0.6787,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012216525254638618,
+      "loss": 0.7455,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012210445348605954,
+      "loss": 0.584,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012204364583380164,
+      "loss": 0.7009,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012198282961324809,
+      "loss": 0.7655,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012192200484803811,
+      "loss": 0.8022,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012186117156181401,
+      "loss": 0.7358,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012180032977822149,
+      "loss": 0.6863,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001217394795209096,
+      "loss": 0.7895,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012167862081353062,
+      "loss": 0.7278,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012161775367974017,
+      "loss": 0.7659,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012155687814319705,
+      "loss": 0.7985,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012149599422756347,
+      "loss": 0.5713,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012143510195650471,
+      "loss": 0.7402,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012137420135368951,
+      "loss": 0.7335,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012131329244278968,
+      "loss": 0.6597,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012125237524748037,
+      "loss": 0.7647,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012119144979143984,
+      "loss": 0.7935,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012113051609834964,
+      "loss": 0.7929,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012106957419189453,
+      "loss": 0.7864,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012100862409576243,
+      "loss": 0.776,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012094766583364445,
+      "loss": 0.7557,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012088669942923485,
+      "loss": 0.7661,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012082572490623112,
+      "loss": 0.7999,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012076474228833384,
+      "loss": 0.7682,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012070375159924673,
+      "loss": 0.8526,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012064275286267674,
+      "loss": 0.8516,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012058174610233384,
+      "loss": 0.6625,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012052073134193118,
+      "loss": 0.6856,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012045970860518503,
+      "loss": 0.6529,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012039867791581469,
+      "loss": 0.7084,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012033763929754265,
+      "loss": 0.6334,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012027659277409435,
+      "loss": 0.6679,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012021553836919848,
+      "loss": 0.7122,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012015447610658661,
+      "loss": 0.663,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012009340600999348,
+      "loss": 0.6929,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012003232810315687,
+      "loss": 0.6253,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011997124240981754,
+      "loss": 0.7906,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011991014895371928,
+      "loss": 0.7594,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011984904775860897,
+      "loss": 0.7518,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011978793884823643,
+      "loss": 0.6396,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011972682224635454,
+      "loss": 0.7874,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011966569797671909,
+      "loss": 0.7477,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011960456606308887,
+      "loss": 0.8189,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001195434265292257,
+      "loss": 0.6393,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001194822793988943,
+      "loss": 0.749,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011942112469586242,
+      "loss": 0.7547,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011935996244390064,
+      "loss": 0.8597,
+      "step": 2363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011929879266678255,
+      "loss": 0.7317,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011923761538828464,
+      "loss": 0.8464,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011917643063218638,
+      "loss": 0.7474,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011911523842227002,
+      "loss": 0.7316,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001190540387823208,
+      "loss": 0.7966,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011899283173612686,
+      "loss": 0.6773,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011893161730747917,
+      "loss": 0.7514,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011887039552017157,
+      "loss": 0.7341,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011880916639800075,
+      "loss": 0.8125,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011874792996476633,
+      "loss": 0.7288,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001186866862442707,
+      "loss": 0.7649,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011862543526031913,
+      "loss": 0.8362,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011856417703671962,
+      "loss": 0.6939,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011850291159728306,
+      "loss": 0.6719,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001184416389658232,
+      "loss": 0.8782,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011838035916615644,
+      "loss": 0.7781,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011831907222210212,
+      "loss": 0.6714,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001182577781574822,
+      "loss": 0.7649,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011819647699612153,
+      "loss": 0.7326,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001181351687618477,
+      "loss": 0.7204,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011807385347849098,
+      "loss": 0.6515,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011801253116988447,
+      "loss": 0.7915,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001179512018598639,
+      "loss": 0.8383,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011788986557226787,
+      "loss": 0.6329,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011782852233093753,
+      "loss": 0.7381,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011776717215971679,
+      "loss": 0.7181,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011770581508245235,
+      "loss": 0.8332,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011764445112299347,
+      "loss": 0.8156,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011758308030519208,
+      "loss": 0.7674,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011752170265290288,
+      "loss": 0.8467,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011746031818998313,
+      "loss": 0.6769,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011739892694029284,
+      "loss": 0.7667,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011733752892769456,
+      "loss": 0.7674,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011727612417605347,
+      "loss": 0.6536,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011721471270923745,
+      "loss": 0.7513,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001171532945511169,
+      "loss": 0.851,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011709186972556494,
+      "loss": 0.7119,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011703043825645715,
+      "loss": 0.7994,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011696900016767171,
+      "loss": 0.6765,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001169075554830895,
+      "loss": 0.7097,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011684610422659381,
+      "loss": 0.6714,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011678464642207056,
+      "loss": 0.6385,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011672318209340825,
+      "loss": 0.7604,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011666171126449779,
+      "loss": 0.8035,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011660023395923278,
+      "loss": 0.7672,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011653875020150919,
+      "loss": 0.6855,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011647726001522561,
+      "loss": 0.7929,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011641576342428305,
+      "loss": 0.6641,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011635426045258508,
+      "loss": 0.7893,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011629275112403766,
+      "loss": 0.8181,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001162312354625493,
+      "loss": 0.7316,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011616971349203092,
+      "loss": 0.69,
+      "step": 2415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011610818523639597,
+      "loss": 0.7034,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011604665071956022,
+      "loss": 0.7336,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011598510996544201,
+      "loss": 0.7213,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011592356299796197,
+      "loss": 0.8401,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011586200984104323,
+      "loss": 0.5858,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011580045051861137,
+      "loss": 0.7925,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011573888505459422,
+      "loss": 0.7109,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011567731347292216,
+      "loss": 0.8237,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001156157357975278,
+      "loss": 0.7202,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011555415205234623,
+      "loss": 0.6723,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011549256226131482,
+      "loss": 0.6939,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011543096644837334,
+      "loss": 0.7877,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011536936463746395,
+      "loss": 0.6868,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011530775685253102,
+      "loss": 0.7299,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011524614311752127,
+      "loss": 0.7351,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011518452345638382,
+      "loss": 0.7475,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011512289789307002,
+      "loss": 0.7627,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011506126645153354,
+      "loss": 0.6937,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011499962915573035,
+      "loss": 0.677,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001149379860296186,
+      "loss": 0.6598,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011487633709715884,
+      "loss": 0.7191,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011481468238231378,
+      "loss": 0.8141,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011475302190904846,
+      "loss": 0.7682,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011469135570133009,
+      "loss": 0.7869,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011462968378312808,
+      "loss": 0.8095,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011456800617841416,
+      "loss": 0.7908,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011450632291116225,
+      "loss": 0.8003,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011444463400534837,
+      "loss": 0.8479,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011438293948495087,
+      "loss": 0.7756,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011432123937395014,
+      "loss": 0.6933,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011425953369632889,
+      "loss": 0.7584,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011419782247607189,
+      "loss": 0.8392,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011413610573716607,
+      "loss": 0.7584,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001140743835036006,
+      "loss": 0.7339,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011401265579936668,
+      "loss": 0.8302,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011395092264845764,
+      "loss": 0.7324,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011388918407486899,
+      "loss": 0.6885,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011382744010259833,
+      "loss": 0.708,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011376569075564536,
+      "loss": 0.6709,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011370393605801184,
+      "loss": 0.7466,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011364217603370159,
+      "loss": 0.7057,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011358041070672058,
+      "loss": 0.791,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011351864010107678,
+      "loss": 0.7157,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011345686424078027,
+      "loss": 0.7108,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011339508314984306,
+      "loss": 0.8072,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011333329685227934,
+      "loss": 0.7587,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011327150537210516,
+      "loss": 0.7062,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011320970873333876,
+      "loss": 0.7777,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011314790696000025,
+      "loss": 0.7545,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011308610007611178,
+      "loss": 0.7676,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011302428810569754,
+      "loss": 0.7479,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011296247107278361,
+      "loss": 0.7684,
+      "step": 2467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011290064900139807,
+      "loss": 0.8258,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011283882191557098,
+      "loss": 0.6857,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011277698983933431,
+      "loss": 0.6542,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011271515279672205,
+      "loss": 0.7276,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011265331081177002,
+      "loss": 0.6927,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011259146390851598,
+      "loss": 0.6815,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011252961211099966,
+      "loss": 0.7296,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011246775544326267,
+      "loss": 0.7275,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011240589392934848,
+      "loss": 0.6758,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011234402759330249,
+      "loss": 0.8231,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011228215645917188,
+      "loss": 0.8052,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011222028055100584,
+      "loss": 0.7038,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011215839989285535,
+      "loss": 0.8649,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011209651450877314,
+      "loss": 0.7722,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011203462442281394,
+      "loss": 0.7474,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011197272965903421,
+      "loss": 0.6768,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011191083024149225,
+      "loss": 0.7223,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011184892619424813,
+      "loss": 0.8103,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011178701754136381,
+      "loss": 0.819,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011172510430690298,
+      "loss": 0.6955,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011166318651493111,
+      "loss": 0.716,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011160126418951542,
+      "loss": 0.7451,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011153933735472497,
+      "loss": 0.7097,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011147740603463051,
+      "loss": 0.7721,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011141547025330455,
+      "loss": 0.7316,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011135353003482132,
+      "loss": 0.6594,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001112915854032568,
+      "loss": 0.7154,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011122963638268864,
+      "loss": 0.7837,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011116768299719626,
+      "loss": 0.8349,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011110572527086078,
+      "loss": 0.659,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011104376322776492,
+      "loss": 0.6945,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011098179689199315,
+      "loss": 0.7836,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011091982628763155,
+      "loss": 0.71,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011085785143876799,
+      "loss": 0.8497,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001107958723694918,
+      "loss": 0.6813,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001107338891038941,
+      "loss": 0.72,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011067190166606757,
+      "loss": 0.6641,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011060991008010657,
+      "loss": 0.7167,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011054791437010698,
+      "loss": 0.8131,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011048591456016636,
+      "loss": 0.6037,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011042391067438384,
+      "loss": 0.5652,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011036190273686018,
+      "loss": 0.7587,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001102998907716976,
+      "loss": 0.6596,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011023787480299996,
+      "loss": 0.8039,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001101758548548727,
+      "loss": 0.6767,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011011383095142278,
+      "loss": 0.7634,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011005180311675869,
+      "loss": 0.7324,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010998977137499045,
+      "loss": 0.6749,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010992773575022961,
+      "loss": 0.7975,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010986569626658918,
+      "loss": 0.7419,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010980365294818378,
+      "loss": 0.6419,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001097416058191294,
+      "loss": 0.6584,
+      "step": 2519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001096795549035436,
+      "loss": 0.6294,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010961750022554533,
+      "loss": 0.6314,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001095554418092551,
+      "loss": 0.7897,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010949337967879478,
+      "loss": 0.7309,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010943131385828774,
+      "loss": 0.6709,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010936924437185878,
+      "loss": 0.6518,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001093071712436341,
+      "loss": 0.7428,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001092450944977413,
+      "loss": 0.636,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010918301415830945,
+      "loss": 0.703,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000109120930249469,
+      "loss": 0.7199,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010905884279535172,
+      "loss": 0.7478,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010899675182009085,
+      "loss": 0.6724,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010893465734782094,
+      "loss": 0.7511,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001088725594026779,
+      "loss": 0.7727,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010881045800879903,
+      "loss": 0.6704,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010874835319032295,
+      "loss": 0.6777,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010868624497138959,
+      "loss": 0.7786,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010862413337614022,
+      "loss": 0.7424,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010856201842871741,
+      "loss": 0.757,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010849990015326508,
+      "loss": 0.7774,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010843777857392836,
+      "loss": 0.778,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010837565371485374,
+      "loss": 0.6885,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010831352560018897,
+      "loss": 0.8098,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010825139425408304,
+      "loss": 0.7142,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010818925970068616,
+      "loss": 0.6969,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001081271219641499,
+      "loss": 0.829,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010806498106862699,
+      "loss": 0.7599,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001080028370382714,
+      "loss": 0.7091,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001079406898972383,
+      "loss": 0.7883,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010787853966968409,
+      "loss": 0.7181,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010781638637976637,
+      "loss": 0.7009,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010775423005164394,
+      "loss": 0.8196,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010769207070947675,
+      "loss": 0.7917,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010762990837742597,
+      "loss": 0.7616,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010756774307965385,
+      "loss": 0.7988,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010750557484032388,
+      "loss": 0.7705,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010744340368360069,
+      "loss": 0.8511,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010738122963364997,
+      "loss": 0.7008,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010731905271463858,
+      "loss": 0.6985,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010725687295073448,
+      "loss": 0.7181,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010719469036610679,
+      "loss": 0.7576,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010713250498492564,
+      "loss": 0.7324,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010707031683136231,
+      "loss": 0.7386,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010700812592958917,
+      "loss": 0.743,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010694593230377959,
+      "loss": 0.7596,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000106883735978108,
+      "loss": 0.7362,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010682153697675,
+      "loss": 0.8172,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010675933532388209,
+      "loss": 0.7138,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010669713104368189,
+      "loss": 0.7526,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010663492416032799,
+      "loss": 0.7081,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010657271469799998,
+      "loss": 0.702,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010651050268087853,
+      "loss": 0.6382,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001064482881331452,
+      "loss": 0.7177,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010638607107898266,
+      "loss": 0.7838,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010632385154257446,
+      "loss": 0.8054,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010626162954810508,
+      "loss": 0.69,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010619940511976005,
+      "loss": 0.772,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010613717828172585,
+      "loss": 0.6761,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001060749490581898,
+      "loss": 0.6848,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010601271747334025,
+      "loss": 0.8228,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010595048355136635,
+      "loss": 0.7542,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001058882473164583,
+      "loss": 0.8018,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010582600879280709,
+      "loss": 0.7549,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010576376800460464,
+      "loss": 0.6861,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010570152497604378,
+      "loss": 0.6497,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010563927973131816,
+      "loss": 0.8372,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010557703229462228,
+      "loss": 0.8343,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010551478269015153,
+      "loss": 0.7454,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010545253094210219,
+      "loss": 0.6959,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010539027707467126,
+      "loss": 0.7254,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010532802111205666,
+      "loss": 0.7802,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010526576307845706,
+      "loss": 0.7353,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010520350299807196,
+      "loss": 0.7802,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010514124089510167,
+      "loss": 0.7247,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010507897679374727,
+      "loss": 0.6971,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010501671071821064,
+      "loss": 0.9694,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010495444269269435,
+      "loss": 0.7338,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010489217274140184,
+      "loss": 0.6824,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010482990088853721,
+      "loss": 0.771,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010476762715830537,
+      "loss": 0.6971,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010470535157491188,
+      "loss": 0.704,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010464307416256309,
+      "loss": 0.7316,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010458079494546604,
+      "loss": 0.7205,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010451851394782843,
+      "loss": 0.7848,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001044562311938587,
+      "loss": 0.8379,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010439394670776596,
+      "loss": 0.6493,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010433166051376006,
+      "loss": 0.6982,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010426937263605134,
+      "loss": 0.8372,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010420708309885096,
+      "loss": 0.7101,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010414479192637066,
+      "loss": 0.6426,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001040824991428228,
+      "loss": 0.6381,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010402020477242043,
+      "loss": 0.7562,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010395790883937718,
+      "loss": 0.675,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001038956113679072,
+      "loss": 0.8076,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010383331238222537,
+      "loss": 0.7857,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010377101190654717,
+      "loss": 0.6918,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010370870996508849,
+      "loss": 0.6195,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010364640658206599,
+      "loss": 0.7314,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010358410178169673,
+      "loss": 0.8711,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010352179558819843,
+      "loss": 0.716,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010345948802578931,
+      "loss": 0.7845,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010339717911868808,
+      "loss": 0.764,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010333486889111408,
+      "loss": 0.7543,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010327255736728707,
+      "loss": 0.8359,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010321024457142732,
+      "loss": 0.8752,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010314793052775562,
+      "loss": 0.7189,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010308561526049328,
+      "loss": 0.8104,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010302329879386203,
+      "loss": 0.7171,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010296098115208407,
+      "loss": 0.7348,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010289866235938208,
+      "loss": 0.8002,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010283634243997917,
+      "loss": 0.7659,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001027740214180989,
+      "loss": 0.6819,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001027116993179653,
+      "loss": 0.6682,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010264937616380273,
+      "loss": 0.7493,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010258705197983598,
+      "loss": 0.6877,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010252472679029029,
+      "loss": 0.7139,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010246240061939128,
+      "loss": 0.7018,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010240007349136489,
+      "loss": 0.8306,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010233774543043753,
+      "loss": 0.7364,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010227541646083589,
+      "loss": 0.8012,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010221308660678708,
+      "loss": 0.6678,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010215075589251846,
+      "loss": 0.6835,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001020884243422578,
+      "loss": 0.8217,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010202609198023323,
+      "loss": 0.6933,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010196375883067309,
+      "loss": 0.7569,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010190142491780607,
+      "loss": 0.7784,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010183909026586118,
+      "loss": 0.6756,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010177675489906773,
+      "loss": 0.7309,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010171441884165524,
+      "loss": 0.7189,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001016520821178536,
+      "loss": 0.6921,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001015897447518928,
+      "loss": 0.7067,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010152740676800325,
+      "loss": 0.7331,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001014650681904155,
+      "loss": 0.6539,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010140272904336039,
+      "loss": 0.7162,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010134038935106891,
+      "loss": 0.6589,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010127804913777234,
+      "loss": 0.708,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010121570842770209,
+      "loss": 0.6573,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010115336724508983,
+      "loss": 0.8073,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010109102561416732,
+      "loss": 0.6797,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010102868355916661,
+      "loss": 0.6055,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010096634110431989,
+      "loss": 0.8003,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010090399827385942,
+      "loss": 0.7041,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010084165509201764,
+      "loss": 0.6528,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010077931158302718,
+      "loss": 0.7836,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010071696777112076,
+      "loss": 0.7737,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010065462368053127,
+      "loss": 0.7207,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010059227933549163,
+      "loss": 0.7455,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010052993476023485,
+      "loss": 0.6506,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010046758997899408,
+      "loss": 0.8677,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010040524501600259,
+      "loss": 0.787,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010034289989549364,
+      "loss": 0.7802,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001002805546417006,
+      "loss": 0.9465,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010021820927885684,
+      "loss": 0.7627,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010015586383119579,
+      "loss": 0.7957,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010009351832295101,
+      "loss": 0.7283,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010003117277835592,
+      "loss": 0.7776,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99688272216441e-05,
+      "loss": 0.798,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990648167704904e-05,
+      "loss": 0.798,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984413616880422e-05,
+      "loss": 0.6874,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97817907211432e-05,
+      "loss": 0.6683,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971944535829942e-05,
+      "loss": 0.7826,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965710010450637e-05,
+      "loss": 0.7401,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959475498399742e-05,
+      "loss": 0.6895,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953241002100591e-05,
+      "loss": 0.7692,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94700652397652e-05,
+      "loss": 0.6577,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940772066450842e-05,
+      "loss": 0.6866,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934537631946874e-05,
+      "loss": 0.774,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928303222887925e-05,
+      "loss": 0.6679,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922068841697284e-05,
+      "loss": 0.6791,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91583449079824e-05,
+      "loss": 0.7191,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909600172614062e-05,
+      "loss": 0.7226,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903365889568014e-05,
+      "loss": 0.7045,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897131644083337e-05,
+      "loss": 0.6582,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89089743858327e-05,
+      "loss": 0.7735,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884663275491022e-05,
+      "loss": 0.7391,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878429157229793e-05,
+      "loss": 0.6675,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872195086222768e-05,
+      "loss": 0.7624,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86596106489311e-05,
+      "loss": 0.6885,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859727095663962e-05,
+      "loss": 0.6102,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853493180958449e-05,
+      "loss": 0.7474,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847259323199674e-05,
+      "loss": 0.7138,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841025524810724e-05,
+      "loss": 0.7053,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834791788214644e-05,
+      "loss": 0.7318,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828558115834477e-05,
+      "loss": 0.6087,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82232451009323e-05,
+      "loss": 0.8247,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816090973413885e-05,
+      "loss": 0.809,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809857508219395e-05,
+      "loss": 0.6136,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803624116932693e-05,
+      "loss": 0.7584,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797390801976679e-05,
+      "loss": 0.7648,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79115756577422e-05,
+      "loss": 0.7164,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784924410748159e-05,
+      "loss": 0.7257,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778691339321297e-05,
+      "loss": 0.7169,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.772458353916412e-05,
+      "loss": 0.7588,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766225456956248e-05,
+      "loss": 0.7968,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759992650863512e-05,
+      "loss": 0.7888,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753759938060876e-05,
+      "loss": 0.6762,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747527320970972e-05,
+      "loss": 0.8677,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741294802016403e-05,
+      "loss": 0.7013,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735062383619732e-05,
+      "loss": 0.685,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728830068203471e-05,
+      "loss": 0.6382,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72259785819011e-05,
+      "loss": 0.685,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716365756002084e-05,
+      "loss": 0.7139,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710133764061794e-05,
+      "loss": 0.7411,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703901884791593e-05,
+      "loss": 0.6919,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.6976701206138e-05,
+      "loss": 0.6361,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691438473950673e-05,
+      "loss": 0.7337,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.685206947224438e-05,
+      "loss": 0.8133,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678975542857273e-05,
+      "loss": 0.6468,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672744263271298e-05,
+      "loss": 0.597,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666513110888594e-05,
+      "loss": 0.6992,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660282088131193e-05,
+      "loss": 0.6784,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654051197421073e-05,
+      "loss": 0.7049,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647820441180159e-05,
+      "loss": 0.8193,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641589821830327e-05,
+      "loss": 0.7117,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.635359341793404e-05,
+      "loss": 0.6859,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629129003491152e-05,
+      "loss": 0.6755,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622898809345288e-05,
+      "loss": 0.693,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616668761777464e-05,
+      "loss": 0.828,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610438863209283e-05,
+      "loss": 0.7046,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604209116062286e-05,
+      "loss": 0.7193,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59797952275796e-05,
+      "loss": 0.8735,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59175008571772e-05,
+      "loss": 0.6555,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585520807362935e-05,
+      "loss": 0.859,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579291690114903e-05,
+      "loss": 0.7046,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57306273639487e-05,
+      "loss": 0.6668,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566833948623998e-05,
+      "loss": 0.7311,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560605329223405e-05,
+      "loss": 0.6828,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554376880614132e-05,
+      "loss": 0.7765,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54814860521716e-05,
+      "loss": 0.7494,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541920505453398e-05,
+      "loss": 0.7207,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535692583743691e-05,
+      "loss": 0.68,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529464842508813e-05,
+      "loss": 0.6463,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523237284169468e-05,
+      "loss": 0.7582,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517009911146282e-05,
+      "loss": 0.6237,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51078272585982e-05,
+      "loss": 0.7299,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504555730730566e-05,
+      "loss": 0.711,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49832892817894e-05,
+      "loss": 0.7606,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492102320625275e-05,
+      "loss": 0.713,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485875910489834e-05,
+      "loss": 0.7508,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479649700192805e-05,
+      "loss": 0.709,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473423692154299e-05,
+      "loss": 0.6923,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.467197888794338e-05,
+      "loss": 0.7213,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460972292532875e-05,
+      "loss": 0.7567,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454746905789784e-05,
+      "loss": 0.7561,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448521730984848e-05,
+      "loss": 0.8039,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442296770537774e-05,
+      "loss": 0.6721,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436072026868185e-05,
+      "loss": 0.6867,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.429847502395623e-05,
+      "loss": 0.6295,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.423623199539536e-05,
+      "loss": 0.8114,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417399120719295e-05,
+      "loss": 0.7333,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.411175268354174e-05,
+      "loss": 0.7674,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404951644863366e-05,
+      "loss": 0.681,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398728252665978e-05,
+      "loss": 0.7351,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392505094181022e-05,
+      "loss": 0.9234,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386282171827416e-05,
+      "loss": 0.664,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.380059488023995e-05,
+      "loss": 0.6878,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373837045189492e-05,
+      "loss": 0.7626,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36761484574256e-05,
+      "loss": 0.6959,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361392892101736e-05,
+      "loss": 0.7269,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.355171186685482e-05,
+      "loss": 0.7567,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348949731912151e-05,
+      "loss": 0.7649,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342728530200004e-05,
+      "loss": 0.7049,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336507583967204e-05,
+      "loss": 0.6482,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330286895631814e-05,
+      "loss": 0.7429,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.324066467611792e-05,
+      "loss": 0.668,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317846302325e-05,
+      "loss": 0.6729,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311626402189202e-05,
+      "loss": 0.7218,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305406769622046e-05,
+      "loss": 0.7207,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299187407041086e-05,
+      "loss": 0.7609,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29296831686377e-05,
+      "loss": 0.7327,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286749501507439e-05,
+      "loss": 0.6658,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280530963389324e-05,
+      "loss": 0.7576,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274312704926552e-05,
+      "loss": 0.635,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268094728536143e-05,
+      "loss": 0.7409,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261877036635007e-05,
+      "loss": 0.7641,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255659631639935e-05,
+      "loss": 0.7402,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249442515967614e-05,
+      "loss": 0.7615,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243225692034618e-05,
+      "loss": 0.7114,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237009162257407e-05,
+      "loss": 0.6997,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230792929052326e-05,
+      "loss": 0.7549,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224576994835608e-05,
+      "loss": 0.7211,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218361362023364e-05,
+      "loss": 0.7055,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212146033031592e-05,
+      "loss": 0.6373,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205931010276174e-05,
+      "loss": 0.7774,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199716296172863e-05,
+      "loss": 0.6783,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193501893137302e-05,
+      "loss": 0.6651,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187287803585011e-05,
+      "loss": 0.6983,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181074029931385e-05,
+      "loss": 0.6454,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1748605745917e-05,
+      "loss": 0.6752,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168647439981104e-05,
+      "loss": 0.6521,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162434628514627e-05,
+      "loss": 0.7957,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156222142607167e-05,
+      "loss": 0.7953,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150009984673498e-05,
+      "loss": 0.7703,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143798157128261e-05,
+      "loss": 0.76,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13758666238598e-05,
+      "loss": 0.669,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131375502861042e-05,
+      "loss": 0.768,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.125164680967706e-05,
+      "loss": 0.6423,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118954199120097e-05,
+      "loss": 0.7485,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112744059732209e-05,
+      "loss": 0.82,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10653426521791e-05,
+      "loss": 0.74,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100324817990917e-05,
+      "loss": 0.6408,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09411572046483e-05,
+      "loss": 0.8387,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087906975053104e-05,
+      "loss": 0.7006,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081698584169057e-05,
+      "loss": 0.6706,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075490550225872e-05,
+      "loss": 0.6629,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069282875636593e-05,
+      "loss": 0.6832,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063075562814125e-05,
+      "loss": 0.8029,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056868614171227e-05,
+      "loss": 0.745,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050662032120525e-05,
+      "loss": 0.7037,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044455819074493e-05,
+      "loss": 0.6223,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03824997744547e-05,
+      "loss": 0.658,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032044509645641e-05,
+      "loss": 0.7296,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025839418087062e-05,
+      "loss": 0.747,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019634705181623e-05,
+      "loss": 0.6854,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013430373341082e-05,
+      "loss": 0.6683,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00722642497704e-05,
+      "loss": 0.6542,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001022862500959e-05,
+      "loss": 0.6699,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994819688324133e-05,
+      "loss": 0.6038,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988616904857725e-05,
+      "loss": 0.7544,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982414514512732e-05,
+      "loss": 0.8235,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976212519700006e-05,
+      "loss": 0.8816,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970010922830243e-05,
+      "loss": 0.71,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963809726313983e-05,
+      "loss": 0.7676,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957608932561616e-05,
+      "loss": 0.6747,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951408543983364e-05,
+      "loss": 0.7374,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945208562989305e-05,
+      "loss": 0.8063,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939008991989347e-05,
+      "loss": 0.6411,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.932809833393244e-05,
+      "loss": 0.8012,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926611089610592e-05,
+      "loss": 0.7568,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920412763050822e-05,
+      "loss": 0.7853,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.914214856123204e-05,
+      "loss": 0.6404,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908017371236843e-05,
+      "loss": 0.7107,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901820310800686e-05,
+      "loss": 0.7696,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.895623677223511e-05,
+      "loss": 0.685,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889427472913924e-05,
+      "loss": 0.6552,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883231700280376e-05,
+      "loss": 0.7972,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877036361731138e-05,
+      "loss": 0.7464,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870841459674324e-05,
+      "loss": 0.6408,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86464699651787e-05,
+      "loss": 0.7107,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858452974669548e-05,
+      "loss": 0.6376,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85225939653695e-05,
+      "loss": 0.6632,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846066264527503e-05,
+      "loss": 0.706,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839873581048461e-05,
+      "loss": 0.8012,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.833681348506894e-05,
+      "loss": 0.807,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.827489569309705e-05,
+      "loss": 0.7419,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821298245863622e-05,
+      "loss": 0.7961,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815107380575189e-05,
+      "loss": 0.7362,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808916975850778e-05,
+      "loss": 0.6574,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.802727034096579e-05,
+      "loss": 0.7426,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796537557718607e-05,
+      "loss": 0.6981,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790348549122688e-05,
+      "loss": 0.7527,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784160010714471e-05,
+      "loss": 0.8568,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777971944899417e-05,
+      "loss": 0.6916,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771784354082813e-05,
+      "loss": 0.706,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765597240669754e-05,
+      "loss": 0.7303,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759410607065153e-05,
+      "loss": 0.7702,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753224455673734e-05,
+      "loss": 0.7492,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747038788900033e-05,
+      "loss": 0.6976,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740853609148402e-05,
+      "loss": 0.6374,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734668918823002e-05,
+      "loss": 0.7049,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728484720327796e-05,
+      "loss": 0.6957,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72230101606657e-05,
+      "loss": 0.8459,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.716117808442904e-05,
+      "loss": 0.8179,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709935099860194e-05,
+      "loss": 0.6567,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70375289272164e-05,
+      "loss": 0.8603,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697571189430247e-05,
+      "loss": 0.7647,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69138999238882e-05,
+      "loss": 0.6786,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685209303999979e-05,
+      "loss": 0.7746,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679029126666129e-05,
+      "loss": 0.7539,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.672849462789486e-05,
+      "loss": 0.6062,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66667031477207e-05,
+      "loss": 0.7402,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660491685015697e-05,
+      "loss": 0.6592,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654313575921974e-05,
+      "loss": 0.7058,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64813598989232e-05,
+      "loss": 0.7276,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641958929327942e-05,
+      "loss": 0.6618,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635782396629843e-05,
+      "loss": 0.7039,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62960639419882e-05,
+      "loss": 0.6136,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623430924435465e-05,
+      "loss": 0.704,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.617255989740168e-05,
+      "loss": 0.7248,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611081592513102e-05,
+      "loss": 0.7285,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604907735154238e-05,
+      "loss": 0.6785,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598734420063334e-05,
+      "loss": 0.8206,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592561649639942e-05,
+      "loss": 0.6626,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586389426283392e-05,
+      "loss": 0.8474,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580217752392816e-05,
+      "loss": 0.624,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574046630367113e-05,
+      "loss": 0.7206,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567876062604989e-05,
+      "loss": 0.6568,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561706051504917e-05,
+      "loss": 0.7184,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.555536599465165e-05,
+      "loss": 0.6019,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.549367708883776e-05,
+      "loss": 0.7064,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543199382158582e-05,
+      "loss": 0.6863,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53703162168719e-05,
+      "loss": 0.601,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530864429866996e-05,
+      "loss": 0.6911,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.524697809095155e-05,
+      "loss": 0.7325,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518531761768624e-05,
+      "loss": 0.6995,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512366290284119e-05,
+      "loss": 0.6734,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506201397038142e-05,
+      "loss": 0.6444,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500037084426967e-05,
+      "loss": 0.7997,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493873354846647e-05,
+      "loss": 0.7356,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487710210692999e-05,
+      "loss": 0.7057,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481547654361618e-05,
+      "loss": 0.7125,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475385688247877e-05,
+      "loss": 0.7447,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.469224314746903e-05,
+      "loss": 0.6292,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463063536253607e-05,
+      "loss": 0.7264,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456903355162667e-05,
+      "loss": 0.7297,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450743773868521e-05,
+      "loss": 0.7188,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44458479476538e-05,
+      "loss": 0.7592,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438426420247221e-05,
+      "loss": 0.5399,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432268652707786e-05,
+      "loss": 0.64,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426111494540579e-05,
+      "loss": 0.7439,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419954948138868e-05,
+      "loss": 0.791,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413799015895678e-05,
+      "loss": 0.7636,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.407643700203806e-05,
+      "loss": 0.6598,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401489003455803e-05,
+      "loss": 0.7059,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39533492804398e-05,
+      "loss": 0.6757,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389181476360406e-05,
+      "loss": 0.747,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383028650796908e-05,
+      "loss": 0.7655,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.376876453745071e-05,
+      "loss": 0.6822,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370724887596238e-05,
+      "loss": 0.6159,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364573954741494e-05,
+      "loss": 0.7051,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358423657571697e-05,
+      "loss": 0.7713,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352273998477442e-05,
+      "loss": 0.6897,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.346124979849082e-05,
+      "loss": 0.7292,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339976604076723e-05,
+      "loss": 0.7113,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33382887355022e-05,
+      "loss": 0.7269,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327681790659179e-05,
+      "loss": 0.6973,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321535357792946e-05,
+      "loss": 0.6051,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315389577340624e-05,
+      "loss": 0.6298,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309244451691054e-05,
+      "loss": 0.6423,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303099983232831e-05,
+      "loss": 0.6119,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296956174354287e-05,
+      "loss": 0.6172,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290813027443508e-05,
+      "loss": 0.634,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284670544888309e-05,
+      "loss": 0.6342,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.278528729076256e-05,
+      "loss": 0.7417,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272387582394652e-05,
+      "loss": 0.7404,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266247107230547e-05,
+      "loss": 0.75,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260107305970717e-05,
+      "loss": 0.7322,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253968181001688e-05,
+      "loss": 0.7538,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247829734709716e-05,
+      "loss": 0.8208,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241691969480793e-05,
+      "loss": 0.7719,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235554887700656e-05,
+      "loss": 0.6823,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229418491754767e-05,
+      "loss": 0.6497,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22328278402832e-05,
+      "loss": 0.7301,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.217147766906253e-05,
+      "loss": 0.6694,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211013442773217e-05,
+      "loss": 0.7892,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20487981401361e-05,
+      "loss": 0.7235,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198746883011556e-05,
+      "loss": 0.6605,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.192614652150905e-05,
+      "loss": 0.767,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186483123815233e-05,
+      "loss": 0.6563,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180352300387848e-05,
+      "loss": 0.668,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17422218425178e-05,
+      "loss": 0.6909,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168092777789793e-05,
+      "loss": 0.7605,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161964083384357e-05,
+      "loss": 0.7125,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155836103417683e-05,
+      "loss": 0.7689,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149708840271695e-05,
+      "loss": 0.7739,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14358229632804e-05,
+      "loss": 0.6416,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137456473968089e-05,
+      "loss": 0.6781,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131331375572931e-05,
+      "loss": 0.687,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125207003523368e-05,
+      "loss": 0.7514,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119083360199925e-05,
+      "loss": 0.6767,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112960447982849e-05,
+      "loss": 0.6554,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106838269252088e-05,
+      "loss": 0.7411,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100716826387315e-05,
+      "loss": 0.855,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094596121767923e-05,
+      "loss": 0.6687,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088476157773002e-05,
+      "loss": 0.7121,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082356936781366e-05,
+      "loss": 0.7218,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.076238461171536e-05,
+      "loss": 0.5973,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.070120733321746e-05,
+      "loss": 0.72,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06400375560994e-05,
+      "loss": 0.8217,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057887530413763e-05,
+      "loss": 0.7446,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051772060110571e-05,
+      "loss": 0.7015,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045657347077433e-05,
+      "loss": 0.6778,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039543393691116e-05,
+      "loss": 0.7319,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033430202328094e-05,
+      "loss": 0.7228,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027317775364548e-05,
+      "loss": 0.6842,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021206115176357e-05,
+      "loss": 0.5879,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015095224139103e-05,
+      "loss": 0.6918,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008985104628076e-05,
+      "loss": 0.6638,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002875759018251e-05,
+      "loss": 0.6599,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996767189684315e-05,
+      "loss": 0.6782,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990659399000654e-05,
+      "loss": 0.6994,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984552389341341e-05,
+      "loss": 0.697,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978446163080154e-05,
+      "loss": 0.7804,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972340722590563e-05,
+      "loss": 0.7211,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966236070245737e-05,
+      "loss": 0.6946,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960132208418533e-05,
+      "loss": 0.7085,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9540291394815e-05,
+      "loss": 0.7615,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947926865806884e-05,
+      "loss": 0.7303,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941825389766617e-05,
+      "loss": 0.7163,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935724713732328e-05,
+      "loss": 0.807,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929624840075329e-05,
+      "loss": 0.7593,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923525771166619e-05,
+      "loss": 0.637,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917427509376889e-05,
+      "loss": 0.6451,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911330057076514e-05,
+      "loss": 0.7472,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905233416635559e-05,
+      "loss": 0.7006,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899137590423759e-05,
+      "loss": 0.7908,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89304258081055e-05,
+      "loss": 0.6916,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886948390165038e-05,
+      "loss": 0.6362,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880855020856019e-05,
+      "loss": 0.7545,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874762475251965e-05,
+      "loss": 0.7081,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868670755721033e-05,
+      "loss": 0.7049,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86257986463105e-05,
+      "loss": 0.6407,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856489804349531e-05,
+      "loss": 0.7519,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.850400577243658e-05,
+      "loss": 0.6991,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844312185680298e-05,
+      "loss": 0.6947,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.838224632025985e-05,
+      "loss": 0.7312,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83213791864694e-05,
+      "loss": 0.6936,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826052047909041e-05,
+      "loss": 0.7057,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819967022177852e-05,
+      "loss": 0.867,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8138828438186e-05,
+      "loss": 0.6745,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807799515196188e-05,
+      "loss": 0.7448,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801717038675192e-05,
+      "loss": 0.7357,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795635416619841e-05,
+      "loss": 0.7656,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789554651394048e-05,
+      "loss": 0.6261,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783474745361383e-05,
+      "loss": 0.7225,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777395700885091e-05,
+      "loss": 0.8067,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771317520328076e-05,
+      "loss": 0.6763,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.765240206052911e-05,
+      "loss": 0.8607,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75916376042182e-05,
+      "loss": 0.7458,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753088185796711e-05,
+      "loss": 0.7962,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747013484539127e-05,
+      "loss": 0.6961,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740939659010292e-05,
+      "loss": 0.6164,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734866711571082e-05,
+      "loss": 0.6567,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728794644582036e-05,
+      "loss": 0.7701,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722723460403342e-05,
+      "loss": 0.657,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716653161394851e-05,
+      "loss": 0.7686,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710583749916072e-05,
+      "loss": 0.7366,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704515228326168e-05,
+      "loss": 0.723,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698447598983946e-05,
+      "loss": 0.8236,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69238086424788e-05,
+      "loss": 0.7219,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686315026476089e-05,
+      "loss": 0.7272,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.680250088026345e-05,
+      "loss": 0.7033,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674186051256072e-05,
+      "loss": 0.5926,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668122918522342e-05,
+      "loss": 0.6317,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662060692181873e-05,
+      "loss": 0.6814,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655999374591035e-05,
+      "loss": 0.7612,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649938968105846e-05,
+      "loss": 0.6182,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643879475081961e-05,
+      "loss": 0.7998,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637820897874688e-05,
+      "loss": 0.7189,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631763238838978e-05,
+      "loss": 0.6008,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625706500329424e-05,
+      "loss": 0.7601,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619650684700257e-05,
+      "loss": 0.704,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613595794305359e-05,
+      "loss": 0.7502,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607541831498247e-05,
+      "loss": 0.6085,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601488798632075e-05,
+      "loss": 0.6746,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595436698059637e-05,
+      "loss": 0.7409,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589385532133364e-05,
+      "loss": 0.7149,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583335303205327e-05,
+      "loss": 0.6935,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577286013627231e-05,
+      "loss": 0.7295,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571237665750419e-05,
+      "loss": 0.6855,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565190261925859e-05,
+      "loss": 0.545,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55914380450416e-05,
+      "loss": 0.5833,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553098295835561e-05,
+      "loss": 0.7275,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547053738269937e-05,
+      "loss": 0.747,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54101013415678e-05,
+      "loss": 0.7257,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534967485845224e-05,
+      "loss": 0.6958,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528925795684027e-05,
+      "loss": 0.7076,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522885066021575e-05,
+      "loss": 0.7201,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516845299205879e-05,
+      "loss": 0.7404,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510806497584582e-05,
+      "loss": 0.7808,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504768663504942e-05,
+      "loss": 0.6356,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498731799313851e-05,
+      "loss": 0.6066,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492695907357812e-05,
+      "loss": 0.6845,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486660989982963e-05,
+      "loss": 0.6158,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480627049535054e-05,
+      "loss": 0.7245,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474594088359459e-05,
+      "loss": 0.686,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468562108801176e-05,
+      "loss": 0.6107,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462531113204809e-05,
+      "loss": 0.6515,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456501103914593e-05,
+      "loss": 0.7365,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450472083274369e-05,
+      "loss": 0.8011,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444444053627605e-05,
+      "loss": 0.8006,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438417017317369e-05,
+      "loss": 0.7348,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432390976686356e-05,
+      "loss": 0.7044,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426365934076865e-05,
+      "loss": 0.8632,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420341891830813e-05,
+      "loss": 0.7135,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414318852289728e-05,
+      "loss": 0.7039,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408296817794747e-05,
+      "loss": 0.7665,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402275790686611e-05,
+      "loss": 0.7341,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.396255773305681e-05,
+      "loss": 0.7555,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390236767991912e-05,
+      "loss": 0.643,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384218777084874e-05,
+      "loss": 0.6948,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378201802923744e-05,
+      "loss": 0.6924,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372185847847301e-05,
+      "loss": 0.739,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366170914193923e-05,
+      "loss": 0.8857,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3601570043016e-05,
+      "loss": 0.7207,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354144120507918e-05,
+      "loss": 0.7436,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34813226515007e-05,
+      "loss": 0.7084,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342121440564837e-05,
+      "loss": 0.7021,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336111649088616e-05,
+      "loss": 0.7954,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330102893057387e-05,
+      "loss": 0.7898,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324095174806737e-05,
+      "loss": 0.8361,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.31808849667185e-05,
+      "loss": 0.6767,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312082860987501e-05,
+      "loss": 0.7761,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306078270088061e-05,
+      "loss": 0.65,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300074726307496e-05,
+      "loss": 0.6256,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294072231979366e-05,
+      "loss": 0.8591,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288070789436818e-05,
+      "loss": 0.6354,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282070401012595e-05,
+      "loss": 0.7982,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276071069039033e-05,
+      "loss": 0.7543,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.270072795848046e-05,
+      "loss": 0.733,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264075583771148e-05,
+      "loss": 0.6623,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258079435139438e-05,
+      "loss": 0.7085,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.252084352283602e-05,
+      "loss": 0.7249,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246090337533905e-05,
+      "loss": 0.7057,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240097393220201e-05,
+      "loss": 0.7037,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23410552167193e-05,
+      "loss": 0.6907,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228114725218113e-05,
+      "loss": 0.6698,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222125006187356e-05,
+      "loss": 0.6443,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216136366907844e-05,
+      "loss": 0.7168,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21014880970734e-05,
+      "loss": 0.7074,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204162336913188e-05,
+      "loss": 0.7205,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198176950852313e-05,
+      "loss": 0.7421,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192192653851219e-05,
+      "loss": 0.6645,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186209448235975e-05,
+      "loss": 0.6811,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180227336332239e-05,
+      "loss": 0.7427,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174246320465241e-05,
+      "loss": 0.7031,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168266402959778e-05,
+      "loss": 0.7481,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162287586140229e-05,
+      "loss": 0.7212,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156309872330537e-05,
+      "loss": 0.6963,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150333263854229e-05,
+      "loss": 0.6915,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144357763034387e-05,
+      "loss": 0.8159,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13838337219367e-05,
+      "loss": 0.6872,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132410093654304e-05,
+      "loss": 0.7574,
+      "step": 3142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.126437929738084e-05,
+      "loss": 0.6801,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120466882766374e-05,
+      "loss": 0.6738,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114496955060099e-05,
+      "loss": 0.7588,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.108528148939752e-05,
+      "loss": 0.7284,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102560466725387e-05,
+      "loss": 0.7425,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096593910736625e-05,
+      "loss": 0.6608,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090628483292652e-05,
+      "loss": 0.731,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0846641867122e-05,
+      "loss": 0.7122,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078701023313582e-05,
+      "loss": 0.8089,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072738995414654e-05,
+      "loss": 0.7034,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066778105332842e-05,
+      "loss": 0.6879,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060818355385124e-05,
+      "loss": 0.6646,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054859747888037e-05,
+      "loss": 0.8308,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048902285157668e-05,
+      "loss": 0.7617,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042945969509674e-05,
+      "loss": 0.7563,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.036990803259245e-05,
+      "loss": 0.5923,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031036788721139e-05,
+      "loss": 0.7055,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025083928209663e-05,
+      "loss": 0.6563,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.01913222403868e-05,
+      "loss": 0.7323,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013181678521592e-05,
+      "loss": 0.569,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007232293971359e-05,
+      "loss": 0.6839,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00128407270049e-05,
+      "loss": 0.6857,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995337017021037e-05,
+      "loss": 0.6646,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.989391129244606e-05,
+      "loss": 0.6674,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983446411682343e-05,
+      "loss": 0.6044,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977502866644936e-05,
+      "loss": 0.8903,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971560496442627e-05,
+      "loss": 0.6812,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965619303385196e-05,
+      "loss": 0.7584,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959679289781968e-05,
+      "loss": 0.6476,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953740457941803e-05,
+      "loss": 0.6631,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947802810173108e-05,
+      "loss": 0.6073,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941866348783834e-05,
+      "loss": 0.672,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935931076081454e-05,
+      "loss": 0.6771,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929996994372995e-05,
+      "loss": 0.7344,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924064105965018e-05,
+      "loss": 0.6858,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918132413163614e-05,
+      "loss": 0.6375,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912201918274414e-05,
+      "loss": 0.7,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906272623602586e-05,
+      "loss": 0.6955,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900344531452824e-05,
+      "loss": 0.6463,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894417644129367e-05,
+      "loss": 0.68,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888491963935967e-05,
+      "loss": 0.6661,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882567493175921e-05,
+      "loss": 0.6559,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876644234152051e-05,
+      "loss": 0.8423,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87072218916671e-05,
+      "loss": 0.6946,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86480136052178e-05,
+      "loss": 0.6243,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858881750518669e-05,
+      "loss": 0.7224,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852963361458305e-05,
+      "loss": 0.8893,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847046195641152e-05,
+      "loss": 0.6023,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841130255367197e-05,
+      "loss": 0.6424,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835215542935938e-05,
+      "loss": 0.7194,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82930206064641e-05,
+      "loss": 0.5985,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823389810797167e-05,
+      "loss": 0.5933,
+      "step": 3194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817478795686279e-05,
+      "loss": 0.6366,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811569017611337e-05,
+      "loss": 0.6732,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805660478869458e-05,
+      "loss": 0.7027,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.79975318175727e-05,
+      "loss": 0.6108,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793847128570923e-05,
+      "loss": 0.7624,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787942321606078e-05,
+      "loss": 0.7659,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782038763157914e-05,
+      "loss": 0.7237,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776136455521127e-05,
+      "loss": 0.613,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770235400989926e-05,
+      "loss": 0.7022,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764335601858034e-05,
+      "loss": 0.745,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75843706041868e-05,
+      "loss": 0.802,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752539778964613e-05,
+      "loss": 0.7349,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746643759788083e-05,
+      "loss": 0.6957,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740749005180862e-05,
+      "loss": 0.6241,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734855517434212e-05,
+      "loss": 0.8084,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72896329883892e-05,
+      "loss": 0.7424,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72307235168527e-05,
+      "loss": 0.7422,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.717182678263056e-05,
+      "loss": 0.642,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711294280861573e-05,
+      "loss": 0.726,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705407161769628e-05,
+      "loss": 0.7213,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699521323275519e-05,
+      "loss": 0.7048,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69363676766706e-05,
+      "loss": 0.7492,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687753497231551e-05,
+      "loss": 0.7293,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681871514255807e-05,
+      "loss": 0.7551,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67599082102613e-05,
+      "loss": 0.6639,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670111419828335e-05,
+      "loss": 0.8132,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664233312947721e-05,
+      "loss": 0.7589,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65835650266909e-05,
+      "loss": 0.7758,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652480991276742e-05,
+      "loss": 0.7649,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646606781054468e-05,
+      "loss": 0.7952,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64073387428556e-05,
+      "loss": 0.7765,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63486227325279e-05,
+      "loss": 0.7313,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628991980238435e-05,
+      "loss": 0.7345,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623122997524257e-05,
+      "loss": 0.7099,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617255327391514e-05,
+      "loss": 0.7241,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611388972120949e-05,
+      "loss": 0.7,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6055239339928e-05,
+      "loss": 0.5712,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599660215286782e-05,
+      "loss": 0.6571,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593797818282112e-05,
+      "loss": 0.6737,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.587936745257477e-05,
+      "loss": 0.7336,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582076998491062e-05,
+      "loss": 0.7334,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576218580260529e-05,
+      "loss": 0.8113,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570361492843034e-05,
+      "loss": 0.6461,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564505738515198e-05,
+      "loss": 0.716,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558651319553142e-05,
+      "loss": 0.7329,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552798238232457e-05,
+      "loss": 0.7135,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54694649682822e-05,
+      "loss": 0.7072,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541096097614982e-05,
+      "loss": 0.6128,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535247042866776e-05,
+      "loss": 0.7764,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52939933485711e-05,
+      "loss": 0.7515,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523552975858967e-05,
+      "loss": 0.6535,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517707968144812e-05,
+      "loss": 0.6034,
+      "step": 3246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511864313986582e-05,
+      "loss": 0.6103,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506022015655683e-05,
+      "loss": 0.7578,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500181075423001e-05,
+      "loss": 0.7161,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494341495558894e-05,
+      "loss": 0.7067,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48850327833318e-05,
+      "loss": 0.6942,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482666426015161e-05,
+      "loss": 0.7297,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.476830940873603e-05,
+      "loss": 0.7445,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470996825176739e-05,
+      "loss": 0.7405,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465164081192269e-05,
+      "loss": 0.7414,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.459332711187369e-05,
+      "loss": 0.6733,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45350271742867e-05,
+      "loss": 0.8478,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447674102182274e-05,
+      "loss": 0.706,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441846867713743e-05,
+      "loss": 0.66,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436021016288105e-05,
+      "loss": 0.6652,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43019655016985e-05,
+      "loss": 0.734,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42437347162293e-05,
+      "loss": 0.6701,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41855178291076e-05,
+      "loss": 0.7487,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.412731486296206e-05,
+      "loss": 0.6921,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406912584041605e-05,
+      "loss": 0.6908,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401095078408741e-05,
+      "loss": 0.6697,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395278971658867e-05,
+      "loss": 0.6547,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389464266052679e-05,
+      "loss": 0.6842,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383650963850336e-05,
+      "loss": 0.7164,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377839067311449e-05,
+      "loss": 0.5873,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372028578695086e-05,
+      "loss": 0.8096,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366219500259766e-05,
+      "loss": 0.6079,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360411834263458e-05,
+      "loss": 0.6239,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354605582963587e-05,
+      "loss": 0.6744,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348800748617023e-05,
+      "loss": 0.7124,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342997333480081e-05,
+      "loss": 0.7161,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337195339808535e-05,
+      "loss": 0.6753,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.331394769857601e-05,
+      "loss": 0.7204,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325595625881942e-05,
+      "loss": 0.6483,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31979791013567e-05,
+      "loss": 0.7316,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314001624872335e-05,
+      "loss": 0.6098,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308206772344934e-05,
+      "loss": 0.705,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302413354805911e-05,
+      "loss": 0.7093,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296621374507154e-05,
+      "loss": 0.6927,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290830833699975e-05,
+      "loss": 0.6935,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285041734635148e-05,
+      "loss": 0.7463,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279254079562873e-05,
+      "loss": 0.6816,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273467870732797e-05,
+      "loss": 0.746,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267683110393996e-05,
+      "loss": 0.7438,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261899800794995e-05,
+      "loss": 0.7379,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256117944183743e-05,
+      "loss": 0.6782,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250337542807636e-05,
+      "loss": 0.8772,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244558598913486e-05,
+      "loss": 0.636,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23878111474756e-05,
+      "loss": 0.6761,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233005092555543e-05,
+      "loss": 0.7376,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227230534582563e-05,
+      "loss": 0.5441,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221457443073163e-05,
+      "loss": 0.6668,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215685820271331e-05,
+      "loss": 0.8163,
+      "step": 3298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20991566842048e-05,
+      "loss": 0.7258,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204146989763449e-05,
+      "loss": 0.6285,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198379786542503e-05,
+      "loss": 0.6654,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192614060999336e-05,
+      "loss": 0.61,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186849815375068e-05,
+      "loss": 0.7423,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181087051910242e-05,
+      "loss": 0.6191,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175325772844827e-05,
+      "loss": 0.7574,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169565980418217e-05,
+      "loss": 0.7412,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16380767686922e-05,
+      "loss": 0.7679,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158050864436072e-05,
+      "loss": 0.8341,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152295545356434e-05,
+      "loss": 0.6169,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14654172186737e-05,
+      "loss": 0.6985,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140789396205376e-05,
+      "loss": 0.7577,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135038570606369e-05,
+      "loss": 0.7283,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12928924730567e-05,
+      "loss": 0.6288,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123541428538024e-05,
+      "loss": 0.713,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117795116537593e-05,
+      "loss": 0.6467,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11205031353795e-05,
+      "loss": 0.712,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106307021772079e-05,
+      "loss": 0.6728,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1005652434723824e-05,
+      "loss": 0.6771,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094824980870668e-05,
+      "loss": 0.7105,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089086236198158e-05,
+      "loss": 0.7268,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083349011685486e-05,
+      "loss": 0.6638,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077613309562696e-05,
+      "loss": 0.656,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071879132059233e-05,
+      "loss": 0.5417,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0661464814039527e-05,
+      "loss": 0.6945,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060415359825122e-05,
+      "loss": 0.7346,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054685769550411e-05,
+      "loss": 0.7255,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0489577128068876e-05,
+      "loss": 0.78,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043231191821033e-05,
+      "loss": 0.7924,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0375062088187284e-05,
+      "loss": 0.6307,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0317827660252534e-05,
+      "loss": 0.7394,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0260608656652926e-05,
+      "loss": 0.6824,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020340509962935e-05,
+      "loss": 0.6972,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0146217011416625e-05,
+      "loss": 0.6989,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008904441424359e-05,
+      "loss": 0.6973,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0031887330333025e-05,
+      "loss": 0.7188,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.997474578190171e-05,
+      "loss": 0.7587,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991761979116039e-05,
+      "loss": 0.6828,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986050938031378e-05,
+      "loss": 0.6872,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98034145715605e-05,
+      "loss": 0.604,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974633538709311e-05,
+      "loss": 0.7025,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9689271849098114e-05,
+      "loss": 0.7649,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9632223979755954e-05,
+      "loss": 0.8004,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957519180124097e-05,
+      "loss": 0.6816,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9518175335721326e-05,
+      "loss": 0.7366,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946117460535918e-05,
+      "loss": 0.6349,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9404189632310515e-05,
+      "loss": 0.6993,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9347220438725236e-05,
+      "loss": 0.6791,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929026704674707e-05,
+      "loss": 0.6631,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923332947851365e-05,
+      "loss": 0.7272,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9176407756156406e-05,
+      "loss": 0.6254,
+      "step": 3350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9119501901800666e-05,
+      "loss": 0.7128,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9062611937565485e-05,
+      "loss": 0.7162,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900573788556384e-05,
+      "loss": 0.725,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894887976790251e-05,
+      "loss": 0.5722,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8892037606682096e-05,
+      "loss": 0.7,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88352114239969e-05,
+      "loss": 0.6636,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877840124193513e-05,
+      "loss": 0.7476,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87216070825787e-05,
+      "loss": 0.6461,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866482896800336e-05,
+      "loss": 0.7415,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8608066920278535e-05,
+      "loss": 0.6634,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8551320961467496e-05,
+      "loss": 0.6919,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849459111362716e-05,
+      "loss": 0.5953,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8437877398808274e-05,
+      "loss": 0.6055,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83811798390553e-05,
+      "loss": 0.6948,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832449845640634e-05,
+      "loss": 0.7996,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8267833272893356e-05,
+      "loss": 0.6368,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8211184310541834e-05,
+      "loss": 0.6181,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8154551591371177e-05,
+      "loss": 0.7088,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809793513739421e-05,
+      "loss": 0.6886,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804133497061759e-05,
+      "loss": 0.7611,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798475111304168e-05,
+      "loss": 0.6814,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792818358666039e-05,
+      "loss": 0.5934,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787163241346141e-05,
+      "loss": 0.8174,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7815097615425937e-05,
+      "loss": 0.6555,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775857921452888e-05,
+      "loss": 0.7711,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7702077232738804e-05,
+      "loss": 0.7444,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764559169201783e-05,
+      "loss": 0.6603,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75891226143217e-05,
+      "loss": 0.7149,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753267002159972e-05,
+      "loss": 0.7196,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747623393579493e-05,
+      "loss": 0.6872,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741981437884381e-05,
+      "loss": 0.7228,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736341137267642e-05,
+      "loss": 0.6313,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730702493921652e-05,
+      "loss": 0.6931,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72506551003813e-05,
+      "loss": 0.7082,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719430187808152e-05,
+      "loss": 0.7472,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713796529422153e-05,
+      "loss": 0.7396,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708164537069911e-05,
+      "loss": 0.6106,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.702534212940574e-05,
+      "loss": 0.6916,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6969055592226226e-05,
+      "loss": 0.7164,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6912785781039046e-05,
+      "loss": 0.7827,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685653271771605e-05,
+      "loss": 0.7034,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680029642412259e-05,
+      "loss": 0.8268,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674407692211766e-05,
+      "loss": 0.721,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668787423355344e-05,
+      "loss": 0.6247,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663168838027587e-05,
+      "loss": 0.7883,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6575519384124155e-05,
+      "loss": 0.7749,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651936726693099e-05,
+      "loss": 0.6895,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6463232050522574e-05,
+      "loss": 0.6206,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640711375671843e-05,
+      "loss": 0.616,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6351012407331635e-05,
+      "loss": 0.6608,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629492802416858e-05,
+      "loss": 0.6809,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6238860629029054e-05,
+      "loss": 0.6165,
+      "step": 3402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.618281024370632e-05,
+      "loss": 0.592,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612677688998691e-05,
+      "loss": 0.7724,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6070760589650916e-05,
+      "loss": 0.7074,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601476136447165e-05,
+      "loss": 0.5489,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5958779236215784e-05,
+      "loss": 0.675,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590281422664349e-05,
+      "loss": 0.6806,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5846866357508086e-05,
+      "loss": 0.8143,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579093565055648e-05,
+      "loss": 0.6425,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5735022127528604e-05,
+      "loss": 0.6149,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567912581015796e-05,
+      "loss": 0.6966,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562324672017125e-05,
+      "loss": 0.8024,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556738487928845e-05,
+      "loss": 0.7514,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5511540309222965e-05,
+      "loss": 0.6942,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5455713031681364e-05,
+      "loss": 0.7548,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539990306836349e-05,
+      "loss": 0.7706,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534411044096259e-05,
+      "loss": 0.6795,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528833517116504e-05,
+      "loss": 0.7067,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52325772806505e-05,
+      "loss": 0.6956,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5176836791091844e-05,
+      "loss": 0.7188,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5121113724155326e-05,
+      "loss": 0.7617,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506540810150027e-05,
+      "loss": 0.7297,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5009719944779236e-05,
+      "loss": 0.6311,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.495404927563812e-05,
+      "loss": 0.7308,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489839611571585e-05,
+      "loss": 0.7714,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484276048664476e-05,
+      "loss": 0.6733,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4787142410050084e-05,
+      "loss": 0.5722,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.47315419075505e-05,
+      "loss": 0.6819,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4675959000757726e-05,
+      "loss": 0.7554,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.462039371127661e-05,
+      "loss": 0.819,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456484606070529e-05,
+      "loss": 0.7464,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450931607063492e-05,
+      "loss": 0.8477,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445380376264979e-05,
+      "loss": 0.866,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439830915832744e-05,
+      "loss": 0.6979,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434283227923842e-05,
+      "loss": 0.7954,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4287373146946383e-05,
+      "loss": 0.7318,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423193178300813e-05,
+      "loss": 0.694,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417650820897357e-05,
+      "loss": 0.6573,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4121102446385684e-05,
+      "loss": 0.6243,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4065714516780444e-05,
+      "loss": 0.6487,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4010344441687045e-05,
+      "loss": 0.6977,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395499224262763e-05,
+      "loss": 0.5784,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389965794111742e-05,
+      "loss": 0.6902,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384434155866467e-05,
+      "loss": 0.7185,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378904311677067e-05,
+      "loss": 0.7407,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373376263692982e-05,
+      "loss": 0.692,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3678500140629375e-05,
+      "loss": 0.6921,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362325564934978e-05,
+      "loss": 0.7594,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356802918456437e-05,
+      "loss": 0.7182,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3512820767739444e-05,
+      "loss": 0.7098,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3457630420334405e-05,
+      "loss": 0.7009,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340245816380156e-05,
+      "loss": 0.7841,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334730401958616e-05,
+      "loss": 0.6428,
+      "step": 3454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3292168009126465e-05,
+      "loss": 0.6645,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3237050153853617e-05,
+      "loss": 0.7804,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318195047519184e-05,
+      "loss": 0.6603,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3126868994558135e-05,
+      "loss": 0.7465,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307180573336256e-05,
+      "loss": 0.6811,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3016760713007996e-05,
+      "loss": 0.7472,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296173395489028e-05,
+      "loss": 0.6605,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2906725480398125e-05,
+      "loss": 0.6638,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2851735310913116e-05,
+      "loss": 0.8081,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279676346780984e-05,
+      "loss": 0.7518,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274180997245564e-05,
+      "loss": 0.8232,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268687484621073e-05,
+      "loss": 0.8043,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2631958110428294e-05,
+      "loss": 0.6345,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25770597864542e-05,
+      "loss": 0.6723,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2522179895627396e-05,
+      "loss": 0.6891,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246731845927937e-05,
+      "loss": 0.6755,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241247549873467e-05,
+      "loss": 0.6856,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235765103531058e-05,
+      "loss": 0.7224,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230284509031715e-05,
+      "loss": 0.7155,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224805768505734e-05,
+      "loss": 0.7072,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219328884082677e-05,
+      "loss": 0.653,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2138538578914e-05,
+      "loss": 0.6146,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208380692060023e-05,
+      "loss": 0.7045,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202909388715951e-05,
+      "loss": 0.6683,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1974399499858597e-05,
+      "loss": 0.7417,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1919723779957e-05,
+      "loss": 0.7161,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186506674870708e-05,
+      "loss": 0.7097,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181042842735381e-05,
+      "loss": 0.6911,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175580883713489e-05,
+      "loss": 0.7988,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170120799928086e-05,
+      "loss": 0.6415,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164662593501484e-05,
+      "loss": 0.8626,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159206266555278e-05,
+      "loss": 0.6749,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1537518212103116e-05,
+      "loss": 0.7469,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148299259586723e-05,
+      "loss": 0.7472,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142848583803901e-05,
+      "loss": 0.657,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137399795980502e-05,
+      "loss": 0.8046,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131952898234461e-05,
+      "loss": 0.7227,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1265078926829656e-05,
+      "loss": 0.7466,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121064781442468e-05,
+      "loss": 0.751,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115623566628698e-05,
+      "loss": 0.6656,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.110184250356633e-05,
+      "loss": 0.7234,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104746834740518e-05,
+      "loss": 0.6842,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099311321893856e-05,
+      "loss": 0.6711,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09387771392942e-05,
+      "loss": 0.8122,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088446012959236e-05,
+      "loss": 0.6598,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083016221094582e-05,
+      "loss": 0.7716,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077588340446009e-05,
+      "loss": 0.6636,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072162373123315e-05,
+      "loss": 0.7095,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066738321235553e-05,
+      "loss": 0.7118,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0613161868910384e-05,
+      "loss": 0.6753,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055895972197333e-05,
+      "loss": 0.6851,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050477679261264e-05,
+      "loss": 0.7391,
+      "step": 3506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0450613101888966e-05,
+      "loss": 0.7188,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0396468670855654e-05,
+      "loss": 0.5983,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0342343520558435e-05,
+      "loss": 0.7009,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028823767203552e-05,
+      "loss": 0.828,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02341511463178e-05,
+      "loss": 0.7622,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018008396442847e-05,
+      "loss": 0.6388,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0126036147383285e-05,
+      "loss": 0.6276,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007200771619045e-05,
+      "loss": 0.7733,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001799869185061e-05,
+      "loss": 0.6366,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996400909535698e-05,
+      "loss": 0.7523,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991003894769508e-05,
+      "loss": 0.6884,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985608826984299e-05,
+      "loss": 0.6374,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980215708277116e-05,
+      "loss": 0.7178,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9748245407442454e-05,
+      "loss": 0.6788,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969435326481217e-05,
+      "loss": 0.7718,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964048067582797e-05,
+      "loss": 0.7624,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958662766143005e-05,
+      "loss": 0.6648,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953279424255083e-05,
+      "loss": 0.7087,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9478980440115254e-05,
+      "loss": 0.6873,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942518627504053e-05,
+      "loss": 0.7167,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937141176823629e-05,
+      "loss": 0.6485,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931765694060457e-05,
+      "loss": 0.6942,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9263921813039596e-05,
+      "loss": 0.6624,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9210206406428114e-05,
+      "loss": 0.668,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.915651074164913e-05,
+      "loss": 0.596,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910283483957393e-05,
+      "loss": 0.7243,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9049178721066244e-05,
+      "loss": 0.6291,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8995542406981954e-05,
+      "loss": 0.6546,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894192591816942e-05,
+      "loss": 0.6539,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888832927546914e-05,
+      "loss": 0.6772,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.883475249971396e-05,
+      "loss": 0.6192,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878119561172904e-05,
+      "loss": 0.8174,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8727658632331694e-05,
+      "loss": 0.5541,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867414158233167e-05,
+      "loss": 0.6795,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862064448253084e-05,
+      "loss": 0.6255,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856716735372332e-05,
+      "loss": 0.6328,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851371021669556e-05,
+      "loss": 0.8065,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846027309222614e-05,
+      "loss": 0.7745,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8406856001085976e-05,
+      "loss": 0.6554,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8353458964038e-05,
+      "loss": 0.6615,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830008200183758e-05,
+      "loss": 0.6975,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.824672513523212e-05,
+      "loss": 0.6344,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8193388384961234e-05,
+      "loss": 0.6774,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8140071771756835e-05,
+      "loss": 0.6125,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808677531634288e-05,
+      "loss": 0.6794,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803349903943548e-05,
+      "loss": 0.7984,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798024296174306e-05,
+      "loss": 0.6637,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7927007103966026e-05,
+      "loss": 0.7053,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7873791486796995e-05,
+      "loss": 0.7092,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782059613092068e-05,
+      "loss": 0.6075,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7767421057014015e-05,
+      "loss": 0.7413,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771426628574595e-05,
+      "loss": 0.5655,
+      "step": 3558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.766113183777753e-05,
+      "loss": 0.8107,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760801773376203e-05,
+      "loss": 0.7184,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7554923994344666e-05,
+      "loss": 0.7027,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7501850640162914e-05,
+      "loss": 0.7612,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74487976918461e-05,
+      "loss": 0.8032,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739576517001575e-05,
+      "loss": 0.6618,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7342753095285496e-05,
+      "loss": 0.7187,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728976148826091e-05,
+      "loss": 0.6177,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723679036953972e-05,
+      "loss": 0.6365,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71838397597116e-05,
+      "loss": 0.7011,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7130909679358246e-05,
+      "loss": 0.6366,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7078000149053505e-05,
+      "loss": 0.7747,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702511118936308e-05,
+      "loss": 0.7426,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697224282084477e-05,
+      "loss": 0.7573,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691939506404828e-05,
+      "loss": 0.5914,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6866567939515446e-05,
+      "loss": 0.6303,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681376146777999e-05,
+      "loss": 0.741,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6760975669367556e-05,
+      "loss": 0.8269,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670821056479592e-05,
+      "loss": 0.6432,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665546617457464e-05,
+      "loss": 0.7858,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660274251920532e-05,
+      "loss": 0.7248,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6550039619181464e-05,
+      "loss": 0.6925,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649735749498849e-05,
+      "loss": 0.6122,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644469616710386e-05,
+      "loss": 0.5842,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639205565599678e-05,
+      "loss": 0.6241,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6339435982128524e-05,
+      "loss": 0.6415,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628683716595216e-05,
+      "loss": 0.5913,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623425922791263e-05,
+      "loss": 0.8101,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618170218844692e-05,
+      "loss": 0.7156,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612916606798371e-05,
+      "loss": 0.7262,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607665088694365e-05,
+      "loss": 0.6483,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.602415666573922e-05,
+      "loss": 0.6646,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5971683424774716e-05,
+      "loss": 0.6558,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5919231184446395e-05,
+      "loss": 0.825,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.58667999651422e-05,
+      "loss": 0.7713,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5814389787242074e-05,
+      "loss": 0.7683,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5762000671117635e-05,
+      "loss": 0.791,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570963263713236e-05,
+      "loss": 0.6782,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565728570564156e-05,
+      "loss": 0.6667,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560495989699226e-05,
+      "loss": 0.6969,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555265523152343e-05,
+      "loss": 0.733,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5500371729565696e-05,
+      "loss": 0.6316,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544810941144143e-05,
+      "loss": 0.7258,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5395868297464916e-05,
+      "loss": 0.6786,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534364840794204e-05,
+      "loss": 0.7806,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529144976317061e-05,
+      "loss": 0.7687,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.523927238343993e-05,
+      "loss": 0.6599,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518711628903131e-05,
+      "loss": 0.6431,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51349815002176e-05,
+      "loss": 0.6329,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5082868037263406e-05,
+      "loss": 0.7728,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5030775920425114e-05,
+      "loss": 0.6386,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497870516995073e-05,
+      "loss": 0.718,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4926655806080044e-05,
+      "loss": 0.6774,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487462784904444e-05,
+      "loss": 0.7004,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482262131906703e-05,
+      "loss": 0.6264,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477063623636257e-05,
+      "loss": 0.826,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471867262113749e-05,
+      "loss": 0.6729,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466673049358992e-05,
+      "loss": 0.7061,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4614809873909604e-05,
+      "loss": 0.6297,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456291078227787e-05,
+      "loss": 0.7459,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451103323886778e-05,
+      "loss": 0.6881,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4459177263843946e-05,
+      "loss": 0.5895,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4407342877362675e-05,
+      "loss": 0.6921,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435553009957174e-05,
+      "loss": 0.7512,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4303738950610675e-05,
+      "loss": 0.63,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425196945061052e-05,
+      "loss": 0.7697,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420022161969386e-05,
+      "loss": 0.6861,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4148495477975006e-05,
+      "loss": 0.721,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40967910455597e-05,
+      "loss": 0.6308,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404510834254526e-05,
+      "loss": 0.6906,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399344738902066e-05,
+      "loss": 0.6626,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3941808205066306e-05,
+      "loss": 0.8434,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3890190810754195e-05,
+      "loss": 0.6687,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3838595226147806e-05,
+      "loss": 0.7614,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378702147130226e-05,
+      "loss": 0.6606,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373546956626407e-05,
+      "loss": 0.6316,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368393953107125e-05,
+      "loss": 0.6102,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3632431385753445e-05,
+      "loss": 0.7247,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358094515033169e-05,
+      "loss": 0.6366,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3529480844818494e-05,
+      "loss": 0.786,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347803848921789e-05,
+      "loss": 0.7429,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3426618103525306e-05,
+      "loss": 0.7604,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337521970772778e-05,
+      "loss": 0.6942,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332384332180362e-05,
+      "loss": 0.7522,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327248896572271e-05,
+      "loss": 0.748,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322115665944634e-05,
+      "loss": 0.6742,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316984642292715e-05,
+      "loss": 0.6656,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3118558276109345e-05,
+      "loss": 0.6994,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3067292238928436e-05,
+      "loss": 0.6909,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3016048331311354e-05,
+      "loss": 0.7603,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2964826573176465e-05,
+      "loss": 0.5975,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291362698443344e-05,
+      "loss": 0.713,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28624495849835e-05,
+      "loss": 0.5491,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281129439471906e-05,
+      "loss": 0.6974,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276016143352404e-05,
+      "loss": 0.7271,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2709050721273624e-05,
+      "loss": 0.6062,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265796227783441e-05,
+      "loss": 0.6126,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2606896123064274e-05,
+      "loss": 0.5777,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255585227681247e-05,
+      "loss": 0.6849,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250483075891964e-05,
+      "loss": 0.632,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245383158921764e-05,
+      "loss": 0.7177,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240285478752967e-05,
+      "loss": 0.5911,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23519003736703e-05,
+      "loss": 0.6358,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2300968367445304e-05,
+      "loss": 0.7059,
+      "step": 3662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225005878865185e-05,
+      "loss": 0.7151,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2199171657078295e-05,
+      "loss": 0.6952,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2148306992504304e-05,
+      "loss": 0.789,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2097464814700815e-05,
+      "loss": 0.8624,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204664514342999e-05,
+      "loss": 0.5904,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1995847998445335e-05,
+      "loss": 0.6257,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194507339949149e-05,
+      "loss": 0.711,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1894321366304427e-05,
+      "loss": 0.7105,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184359191861128e-05,
+      "loss": 0.6083,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1792885076130417e-05,
+      "loss": 0.7314,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174220085857145e-05,
+      "loss": 0.6546,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169153928563511e-05,
+      "loss": 0.665,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164090037701347e-05,
+      "loss": 0.7199,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159028415238968e-05,
+      "loss": 0.7016,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1539690631438075e-05,
+      "loss": 0.7535,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1489119833824276e-05,
+      "loss": 0.7088,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1438571779204914e-05,
+      "loss": 0.7454,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138804648722795e-05,
+      "loss": 0.8008,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133754397753229e-05,
+      "loss": 0.7779,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1287064269748174e-05,
+      "loss": 0.6828,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123660738349691e-05,
+      "loss": 0.716,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118617333839086e-05,
+      "loss": 0.7981,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113576215403366e-05,
+      "loss": 0.7264,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.108537385001995e-05,
+      "loss": 0.6606,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103500844593546e-05,
+      "loss": 0.6466,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098466596135715e-05,
+      "loss": 0.6675,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093434641585293e-05,
+      "loss": 0.7153,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0884049828981866e-05,
+      "loss": 0.6194,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083377622029403e-05,
+      "loss": 0.6802,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078352560933069e-05,
+      "loss": 0.6849,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073329801562408e-05,
+      "loss": 0.6893,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068309345869745e-05,
+      "loss": 0.7958,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0632911958065237e-05,
+      "loss": 0.6576,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05827535332328e-05,
+      "loss": 0.7506,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053261820369655e-05,
+      "loss": 0.6321,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048250598894393e-05,
+      "loss": 0.6599,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043241690845337e-05,
+      "loss": 0.6707,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038235098169442e-05,
+      "loss": 0.7794,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033230822812745e-05,
+      "loss": 0.6677,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0282288667204017e-05,
+      "loss": 0.7287,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023229231836652e-05,
+      "loss": 0.6721,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0182319201048334e-05,
+      "loss": 0.6989,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013236933467395e-05,
+      "loss": 0.6165,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.008244273865866e-05,
+      "loss": 0.7255,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00325394324088e-05,
+      "loss": 0.6835,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998265943532161e-05,
+      "loss": 0.7844,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993280276678526e-05,
+      "loss": 0.7159,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988296944617894e-05,
+      "loss": 0.7077,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9833159492872654e-05,
+      "loss": 0.717,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9783372926227424e-05,
+      "loss": 0.7223,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9733609765595113e-05,
+      "loss": 0.7324,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968387003031849e-05,
+      "loss": 0.6247,
+      "step": 3714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963415373973124e-05,
+      "loss": 0.6434,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9584460913157886e-05,
+      "loss": 0.6297,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953479156991396e-05,
+      "loss": 0.6258,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948514572930568e-05,
+      "loss": 0.7121,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943552341063034e-05,
+      "loss": 0.787,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9385924633175896e-05,
+      "loss": 0.6447,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933634941622124e-05,
+      "loss": 0.6585,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9286797779036154e-05,
+      "loss": 0.7106,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9237269740881165e-05,
+      "loss": 0.6437,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9187765321007674e-05,
+      "loss": 0.7447,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913828453865789e-05,
+      "loss": 0.6952,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9088827413064797e-05,
+      "loss": 0.6341,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903939396345231e-05,
+      "loss": 0.6169,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8989984209034966e-05,
+      "loss": 0.7155,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894059816901826e-05,
+      "loss": 0.7726,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889123586259835e-05,
+      "loss": 0.6812,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88418973089622e-05,
+      "loss": 0.5881,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879258252728758e-05,
+      "loss": 0.6261,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874329153674293e-05,
+      "loss": 0.7525,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8694024356487577e-05,
+      "loss": 0.5789,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8644781005671497e-05,
+      "loss": 0.6189,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859556150343536e-05,
+      "loss": 0.6801,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854636586891073e-05,
+      "loss": 0.7482,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849719412121972e-05,
+      "loss": 0.7427,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844804627947528e-05,
+      "loss": 0.7631,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839892236278102e-05,
+      "loss": 0.6758,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8349822390231227e-05,
+      "loss": 0.8566,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830074638091092e-05,
+      "loss": 0.6628,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825169435389574e-05,
+      "loss": 0.7076,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8202666328252135e-05,
+      "loss": 0.6328,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815366232303711e-05,
+      "loss": 0.7307,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810468235729833e-05,
+      "loss": 0.6968,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8055726450074216e-05,
+      "loss": 0.7,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800679462039375e-05,
+      "loss": 0.6999,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795788688727657e-05,
+      "loss": 0.6586,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7909003269732926e-05,
+      "loss": 0.7206,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78601437867638e-05,
+      "loss": 0.6098,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.781130845736068e-05,
+      "loss": 0.6471,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7762497300505675e-05,
+      "loss": 0.6549,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771371033517159e-05,
+      "loss": 0.6402,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.766494758032171e-05,
+      "loss": 0.8094,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.761620905491005e-05,
+      "loss": 0.7411,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756749477788101e-05,
+      "loss": 0.6154,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75188047681697e-05,
+      "loss": 0.7223,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7470139044701824e-05,
+      "loss": 0.6768,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742149762639354e-05,
+      "loss": 0.7787,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7372880532151676e-05,
+      "loss": 0.7968,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.73242877808735e-05,
+      "loss": 0.6319,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.727571939144685e-05,
+      "loss": 0.7619,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7227175382750147e-05,
+      "loss": 0.5452,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.717865577365228e-05,
+      "loss": 0.6994,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7130160583012665e-05,
+      "loss": 0.715,
+      "step": 3766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708168982968118e-05,
+      "loss": 0.6971,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.703324353249833e-05,
+      "loss": 0.6306,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698482171029501e-05,
+      "loss": 0.6948,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.69364243818926e-05,
+      "loss": 0.7471,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.688805156610304e-05,
+      "loss": 0.5656,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6839703281728675e-05,
+      "loss": 0.6445,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679137954756232e-05,
+      "loss": 0.6601,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.674308038238725e-05,
+      "loss": 0.6613,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6694805804977186e-05,
+      "loss": 0.7496,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6646555834096355e-05,
+      "loss": 0.7417,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.659833048849929e-05,
+      "loss": 0.858,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.655012978693113e-05,
+      "loss": 0.6059,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6501953748127284e-05,
+      "loss": 0.6397,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645380239081359e-05,
+      "loss": 0.549,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.640567573370639e-05,
+      "loss": 0.6496,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.635757379551234e-05,
+      "loss": 0.7109,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630949659492853e-05,
+      "loss": 0.7993,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.626144415064239e-05,
+      "loss": 0.5333,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.621341648133172e-05,
+      "loss": 0.6711,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.61654136056648e-05,
+      "loss": 0.7893,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.611743554230013e-05,
+      "loss": 0.6633,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6069482309886694e-05,
+      "loss": 0.618,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602155392706372e-05,
+      "loss": 0.6543,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5973650412460844e-05,
+      "loss": 0.6205,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.592577178469799e-05,
+      "loss": 0.6794,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5877918062385406e-05,
+      "loss": 0.5711,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.583008926412374e-05,
+      "loss": 0.5893,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578228540850388e-05,
+      "loss": 0.6633,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573450651410698e-05,
+      "loss": 0.6869,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5686752599504625e-05,
+      "loss": 0.7192,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563902368325853e-05,
+      "loss": 0.6585,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559131978392086e-05,
+      "loss": 0.7075,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5543640920033925e-05,
+      "loss": 0.5991,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.549598711013035e-05,
+      "loss": 0.6785,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5448358372733016e-05,
+      "loss": 0.7896,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540075472635505e-05,
+      "loss": 0.6233,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535317618949987e-05,
+      "loss": 0.6786,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5305622780661075e-05,
+      "loss": 0.7476,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525809451832257e-05,
+      "loss": 0.6922,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.521059142095841e-05,
+      "loss": 0.6169,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.51631135070329e-05,
+      "loss": 0.7273,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511566079500055e-05,
+      "loss": 0.6948,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5068233303306053e-05,
+      "loss": 0.7214,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502083105038439e-05,
+      "loss": 0.6721,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.497345405466065e-05,
+      "loss": 0.7143,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.492610233455006e-05,
+      "loss": 0.6924,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487877590845817e-05,
+      "loss": 0.6994,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483147479478054e-05,
+      "loss": 0.7475,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478419901190305e-05,
+      "loss": 0.7401,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4736948578201536e-05,
+      "loss": 0.7665,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.468972351204218e-05,
+      "loss": 0.7962,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4642523831781206e-05,
+      "loss": 0.8371,
+      "step": 3818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459534955576491e-05,
+      "loss": 0.6776,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4548200702329866e-05,
+      "loss": 0.6627,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4501077289802665e-05,
+      "loss": 0.705,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4453979336499974e-05,
+      "loss": 0.7131,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4406906860728705e-05,
+      "loss": 0.6966,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.435985988078574e-05,
+      "loss": 0.6262,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4312838414958085e-05,
+      "loss": 0.6976,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4265842481522816e-05,
+      "loss": 0.6592,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4218872098747144e-05,
+      "loss": 0.8066,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.417192728488831e-05,
+      "loss": 0.5766,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.412500805819357e-05,
+      "loss": 0.5978,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.407811443690032e-05,
+      "loss": 0.6698,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.403124643923596e-05,
+      "loss": 0.7892,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.398440408341792e-05,
+      "loss": 0.6871,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393758738765368e-05,
+      "loss": 0.8079,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389079637014071e-05,
+      "loss": 0.6647,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3844031049066594e-05,
+      "loss": 0.6982,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3797291442608804e-05,
+      "loss": 0.7291,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.375057756893494e-05,
+      "loss": 0.6755,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3703889446202494e-05,
+      "loss": 0.7381,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.365722709255898e-05,
+      "loss": 0.7297,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.361059052614194e-05,
+      "loss": 0.6026,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3563979765078866e-05,
+      "loss": 0.6253,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.351739482748719e-05,
+      "loss": 0.7601,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347083573147434e-05,
+      "loss": 0.5543,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3424302495137646e-05,
+      "loss": 0.6502,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.337779513656452e-05,
+      "loss": 0.7499,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.333131367383213e-05,
+      "loss": 0.6552,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3284858125007754e-05,
+      "loss": 0.7205,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323842850814849e-05,
+      "loss": 0.6734,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.319202484130137e-05,
+      "loss": 0.6854,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3145647142503365e-05,
+      "loss": 0.7137,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3099295429781295e-05,
+      "loss": 0.6609,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3052969721152005e-05,
+      "loss": 0.7681,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300667003462208e-05,
+      "loss": 0.6397,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.296039638818812e-05,
+      "loss": 0.6848,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.291414879983652e-05,
+      "loss": 0.7353,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286792728754353e-05,
+      "loss": 0.6832,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2821731869275385e-05,
+      "loss": 0.6874,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.277556256298806e-05,
+      "loss": 0.7826,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2729419386627414e-05,
+      "loss": 0.7294,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268330235812917e-05,
+      "loss": 0.6628,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263721149541881e-05,
+      "loss": 0.8097,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.25911468164118e-05,
+      "loss": 0.7439,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254510833901325e-05,
+      "loss": 0.6621,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249909608111825e-05,
+      "loss": 0.7398,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2453110060611594e-05,
+      "loss": 0.6957,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240715029536787e-05,
+      "loss": 0.6716,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2361216803251523e-05,
+      "loss": 0.8807,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2315309602116715e-05,
+      "loss": 0.7075,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22694287098075e-05,
+      "loss": 0.6697,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22235741441576e-05,
+      "loss": 0.5809,
+      "step": 3870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217774592299051e-05,
+      "loss": 0.7845,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213194406411957e-05,
+      "loss": 0.6937,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.208616858534775e-05,
+      "loss": 0.7221,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2040419504467925e-05,
+      "loss": 0.6111,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199469683926256e-05,
+      "loss": 0.6561,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.194900060750393e-05,
+      "loss": 0.7513,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1903330826954e-05,
+      "loss": 0.6133,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185768751536444e-05,
+      "loss": 0.6312,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.181207069047673e-05,
+      "loss": 0.7219,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1766480370021966e-05,
+      "loss": 0.7597,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1720916571720914e-05,
+      "loss": 0.7211,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1675379313284135e-05,
+      "loss": 0.8249,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1629868612411806e-05,
+      "loss": 0.6858,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.158438448679377e-05,
+      "loss": 0.7186,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.153892695410956e-05,
+      "loss": 0.7801,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149349603202841e-05,
+      "loss": 0.7839,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144809173820916e-05,
+      "loss": 0.6823,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140271409030029e-05,
+      "loss": 0.7481,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1357363105939994e-05,
+      "loss": 0.6005,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131203880275604e-05,
+      "loss": 0.7523,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1266741198365835e-05,
+      "loss": 0.5475,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122147031037642e-05,
+      "loss": 0.6635,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1176226156384406e-05,
+      "loss": 0.6911,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.113100875397612e-05,
+      "loss": 0.6364,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1085818120727364e-05,
+      "loss": 0.6678,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104065427420364e-05,
+      "loss": 0.643,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.099551723195999e-05,
+      "loss": 0.6428,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0950407011540993e-05,
+      "loss": 0.7982,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.090532363048091e-05,
+      "loss": 0.7286,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0860267106303485e-05,
+      "loss": 0.6651,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081523745652204e-05,
+      "loss": 0.8204,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0770234698639435e-05,
+      "loss": 0.6476,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0725258850148145e-05,
+      "loss": 0.6914,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0680309928530124e-05,
+      "loss": 0.7018,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063538795125685e-05,
+      "loss": 0.7489,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.059049293578939e-05,
+      "loss": 0.7032,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0545624899578304e-05,
+      "loss": 0.7011,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050078386006362e-05,
+      "loss": 0.6011,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0455969834674926e-05,
+      "loss": 0.6919,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0411182840831255e-05,
+      "loss": 0.5941,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0366422895941237e-05,
+      "loss": 0.6558,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0321690017402848e-05,
+      "loss": 0.6992,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.02769842226037e-05,
+      "loss": 0.7987,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0232305528920744e-05,
+      "loss": 0.8042,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0187653953720417e-05,
+      "loss": 0.7486,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0143029514358723e-05,
+      "loss": 0.6369,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0098432228180993e-05,
+      "loss": 0.8098,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005386211252207e-05,
+      "loss": 0.8159,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0009319184706196e-05,
+      "loss": 0.7096,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9964803462047053e-05,
+      "loss": 0.785,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9920314961847806e-05,
+      "loss": 0.7022,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9875853701400947e-05,
+      "loss": 0.6968,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9831419697988483e-05,
+      "loss": 0.7406,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9787012968881755e-05,
+      "loss": 0.7515,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974263353134149e-05,
+      "loss": 0.673,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9698281402617855e-05,
+      "loss": 0.6635,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965395659995035e-05,
+      "loss": 0.6271,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9609659140567947e-05,
+      "loss": 0.6154,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9565389041688895e-05,
+      "loss": 0.6604,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9521146320520808e-05,
+      "loss": 0.6872,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9476930994260766e-05,
+      "loss": 0.6595,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9432743080095047e-05,
+      "loss": 0.8004,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9388582595199443e-05,
+      "loss": 0.7655,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9344449556738927e-05,
+      "loss": 0.7617,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9300343981867906e-05,
+      "loss": 0.5747,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.925626588773005e-05,
+      "loss": 0.6884,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9212215291458368e-05,
+      "loss": 0.7445,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.916819221017524e-05,
+      "loss": 0.6846,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9124196660992254e-05,
+      "loss": 0.6882,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9080228661010332e-05,
+      "loss": 0.7357,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9036288227319752e-05,
+      "loss": 0.6588,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899237537699998e-05,
+      "loss": 0.7844,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894849012711982e-05,
+      "loss": 0.7253,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8904632494737283e-05,
+      "loss": 0.6309,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8860802496899765e-05,
+      "loss": 0.6228,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8817000150643812e-05,
+      "loss": 0.6292,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8773225472995225e-05,
+      "loss": 0.7362,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8729478480969153e-05,
+      "loss": 0.7028,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868575919156985e-05,
+      "loss": 0.6595,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8642067621790923e-05,
+      "loss": 0.6074,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8598403788615147e-05,
+      "loss": 0.723,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8554767709014417e-05,
+      "loss": 0.6561,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851115939995005e-05,
+      "loss": 0.6239,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8467578878372392e-05,
+      "loss": 0.7398,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8424026161221096e-05,
+      "loss": 0.6011,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8380501265424963e-05,
+      "loss": 0.7218,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.833700420790193e-05,
+      "loss": 0.7215,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8293535005559235e-05,
+      "loss": 0.5763,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82500936752932e-05,
+      "loss": 0.6334,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8206680233989312e-05,
+      "loss": 0.6969,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8163294698522226e-05,
+      "loss": 0.818,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811993708575581e-05,
+      "loss": 0.7502,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.807660741254302e-05,
+      "loss": 0.7649,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.803330569572591e-05,
+      "loss": 0.7091,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7990031952135797e-05,
+      "loss": 0.7801,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.794678619859301e-05,
+      "loss": 0.6182,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7903568451907037e-05,
+      "loss": 0.6259,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786037872887649e-05,
+      "loss": 0.6567,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7817217046289034e-05,
+      "loss": 0.6544,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777408342092155e-05,
+      "loss": 0.7132,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7730977869539876e-05,
+      "loss": 0.6801,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768790040889906e-05,
+      "loss": 0.8107,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764485105574316e-05,
+      "loss": 0.8113,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7601829826805282e-05,
+      "loss": 0.8013,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7558836738807703e-05,
+      "loss": 0.6181,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7515871808461668e-05,
+      "loss": 0.7312,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.747293505246752e-05,
+      "loss": 0.7298,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7430026487514625e-05,
+      "loss": 0.7421,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7387146130281392e-05,
+      "loss": 0.6768,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7344293997435332e-05,
+      "loss": 0.6651,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7301470105632876e-05,
+      "loss": 0.7292,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725867447151961e-05,
+      "loss": 0.6827,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.721590711173e-05,
+      "loss": 0.7925,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7173168042887608e-05,
+      "loss": 0.7633,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713045728160496e-05,
+      "loss": 0.7034,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7087774844483582e-05,
+      "loss": 0.6967,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.704512074811405e-05,
+      "loss": 0.6777,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7002495009075857e-05,
+      "loss": 0.6894,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695989764393745e-05,
+      "loss": 0.6976,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6917328669256358e-05,
+      "loss": 0.7558,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6874788101578952e-05,
+      "loss": 0.7442,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6832275957440655e-05,
+      "loss": 0.605,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6789792253365797e-05,
+      "loss": 0.6257,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6747337005867657e-05,
+      "loss": 0.7182,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6704910231448442e-05,
+      "loss": 0.7196,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.666251194659929e-05,
+      "loss": 0.7079,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.662014216780033e-05,
+      "loss": 0.6389,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65778009115205e-05,
+      "loss": 0.6565,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6535488194217784e-05,
+      "loss": 0.7286,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649320403233897e-05,
+      "loss": 0.6869,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6450948442319767e-05,
+      "loss": 0.6978,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6408721440584795e-05,
+      "loss": 0.7197,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.636652304354753e-05,
+      "loss": 0.7171,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6324353267610413e-05,
+      "loss": 0.6492,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6282212129164674e-05,
+      "loss": 0.8482,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.624009964459041e-05,
+      "loss": 0.5996,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.619801583025666e-05,
+      "loss": 0.8056,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615596070252123e-05,
+      "loss": 0.6744,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6113934277730865e-05,
+      "loss": 0.6106,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607193657222108e-05,
+      "loss": 0.7036,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602996760231624e-05,
+      "loss": 0.7176,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598802738432956e-05,
+      "loss": 0.6656,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594611593456303e-05,
+      "loss": 0.6781,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5904233269307565e-05,
+      "loss": 0.7273,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5862379404842796e-05,
+      "loss": 0.6987,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5820554357437166e-05,
+      "loss": 0.6621,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577875814334798e-05,
+      "loss": 0.7288,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573699077882127e-05,
+      "loss": 0.7373,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5695252280091884e-05,
+      "loss": 0.7704,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5653542663383423e-05,
+      "loss": 0.6465,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5611861944908334e-05,
+      "loss": 0.8017,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5570210140867744e-05,
+      "loss": 0.7161,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5528587267451554e-05,
+      "loss": 0.717,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5486993340838505e-05,
+      "loss": 0.7174,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5445428377196e-05,
+      "loss": 0.6906,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5403892392680196e-05,
+      "loss": 0.678,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5362385403436008e-05,
+      "loss": 0.6346,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532090742559704e-05,
+      "loss": 0.7281,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527945847528571e-05,
+      "loss": 0.6811,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5238038568613043e-05,
+      "loss": 0.6378,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519664772167888e-05,
+      "loss": 0.8219,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.515528595057167e-05,
+      "loss": 0.6477,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5113953271368586e-05,
+      "loss": 0.619,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5072649700135576e-05,
+      "loss": 0.6043,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503137525292716e-05,
+      "loss": 0.8433,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49901299457866e-05,
+      "loss": 0.6773,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494891379474581e-05,
+      "loss": 0.711,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.490772681582535e-05,
+      "loss": 0.6448,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4866569025034525e-05,
+      "loss": 0.6725,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4825440438371182e-05,
+      "loss": 0.7273,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4784341071821938e-05,
+      "loss": 0.6421,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474327094136195e-05,
+      "loss": 0.6446,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4702230062955057e-05,
+      "loss": 0.6353,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.46612184525537e-05,
+      "loss": 0.7753,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.462023612609895e-05,
+      "loss": 0.6189,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4579283099520554e-05,
+      "loss": 0.815,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4538359388736776e-05,
+      "loss": 0.7656,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4497465009654598e-05,
+      "loss": 0.6038,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4456599978169493e-05,
+      "loss": 0.6854,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4415764310165545e-05,
+      "loss": 0.6734,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4374958021515502e-05,
+      "loss": 0.68,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4334181128080624e-05,
+      "loss": 0.7628,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429343364571075e-05,
+      "loss": 0.6686,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42527155902443e-05,
+      "loss": 0.6675,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4212026977508218e-05,
+      "loss": 0.793,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4171367823318115e-05,
+      "loss": 0.8248,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4130738143478003e-05,
+      "loss": 0.5342,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409013795378058e-05,
+      "loss": 0.6021,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404956727000698e-05,
+      "loss": 0.6452,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400902610792688e-05,
+      "loss": 0.7447,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3968514483298532e-05,
+      "loss": 0.6044,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3928032411868618e-05,
+      "loss": 0.6049,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3887579909372472e-05,
+      "loss": 0.6739,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3847156991533824e-05,
+      "loss": 0.6411,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3806763674064892e-05,
+      "loss": 0.6566,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3766399972666498e-05,
+      "loss": 0.6146,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3726065903027816e-05,
+      "loss": 0.7011,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368576148082665e-05,
+      "loss": 0.6336,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364548672172916e-05,
+      "loss": 0.6773,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360524164139002e-05,
+      "loss": 0.701,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.356502625545236e-05,
+      "loss": 0.7711,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3524840579547757e-05,
+      "loss": 0.7095,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3484684629296306e-05,
+      "loss": 0.7655,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.344455842030646e-05,
+      "loss": 0.7005,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340446196817515e-05,
+      "loss": 0.6855,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3364395288487762e-05,
+      "loss": 0.703,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332435839681808e-05,
+      "loss": 0.7647,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3284351308728327e-05,
+      "loss": 0.6969,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324437403976908e-05,
+      "loss": 0.7444,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3204426605479444e-05,
+      "loss": 0.7543,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3164509021386826e-05,
+      "loss": 0.7244,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3124621303007056e-05,
+      "loss": 0.6625,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3084763465844406e-05,
+      "loss": 0.6304,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3044935525391453e-05,
+      "loss": 0.7446,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3005137497129192e-05,
+      "loss": 0.619,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2965369396527047e-05,
+      "loss": 0.6237,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2925631239042667e-05,
+      "loss": 0.6746,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2885923040122216e-05,
+      "loss": 0.602,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284624481520008e-05,
+      "loss": 0.8053,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2806596579699136e-05,
+      "loss": 0.7215,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2766978349030497e-05,
+      "loss": 0.6904,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2727390138593606e-05,
+      "loss": 0.7666,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2687831963776342e-05,
+      "loss": 0.8152,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2648303839954797e-05,
+      "loss": 0.6621,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2608805782493448e-05,
+      "loss": 0.6656,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256933780674504e-05,
+      "loss": 0.639,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.252989992805068e-05,
+      "loss": 0.6638,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249049216173975e-05,
+      "loss": 0.7317,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2451114523129867e-05,
+      "loss": 0.5848,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2411767027527074e-05,
+      "loss": 0.7617,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2372449690225583e-05,
+      "loss": 0.7157,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2333162526507922e-05,
+      "loss": 0.6746,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2293905551644878e-05,
+      "loss": 0.7726,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225467878089549e-05,
+      "loss": 0.7652,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2215482229507147e-05,
+      "loss": 0.8008,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2176315912715352e-05,
+      "loss": 0.6558,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213717984574398e-05,
+      "loss": 0.8,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2098074043805095e-05,
+      "loss": 0.7443,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.205899852209895e-05,
+      "loss": 0.658,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2019953295814145e-05,
+      "loss": 0.6023,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1980938380127393e-05,
+      "loss": 0.5786,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1941953790203683e-05,
+      "loss": 0.6866,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1902999541196202e-05,
+      "loss": 0.75,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1864075648246296e-05,
+      "loss": 0.648,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1825182126483645e-05,
+      "loss": 0.5538,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1786318991025957e-05,
+      "loss": 0.6649,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1747486256979276e-05,
+      "loss": 0.7277,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1708683939437745e-05,
+      "loss": 0.5977,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1669912053483687e-05,
+      "loss": 0.7476,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1631170614187614e-05,
+      "loss": 0.6426,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1592459636608186e-05,
+      "loss": 0.682,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1553779135792273e-05,
+      "loss": 0.5896,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1515129126774858e-05,
+      "loss": 0.7069,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1476509624579034e-05,
+      "loss": 0.7106,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1437920644216138e-05,
+      "loss": 0.7168,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1399362200685534e-05,
+      "loss": 0.7449,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136083430897482e-05,
+      "loss": 0.7195,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132233698405963e-05,
+      "loss": 0.6434,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128387024090377e-05,
+      "loss": 0.7355,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1245434094459137e-05,
+      "loss": 0.676,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.12070285596657e-05,
+      "loss": 0.7926,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116865365145163e-05,
+      "loss": 0.7251,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1130309384733116e-05,
+      "loss": 0.7672,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1091995774414407e-05,
+      "loss": 0.7539,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1053712835387963e-05,
+      "loss": 0.6183,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1015460582534187e-05,
+      "loss": 0.8014,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0977239030721617e-05,
+      "loss": 0.7645,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0939048194806833e-05,
+      "loss": 0.6783,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090088808963454e-05,
+      "loss": 0.7068,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0862758730037424e-05,
+      "loss": 0.763,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0824660130836217e-05,
+      "loss": 0.7433,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.078659230683978e-05,
+      "loss": 0.6886,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0748555272844905e-05,
+      "loss": 0.7219,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0710549043636517e-05,
+      "loss": 0.6939,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067257363398749e-05,
+      "loss": 0.792,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063462905865875e-05,
+      "loss": 0.7456,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0596715332399242e-05,
+      "loss": 0.6951,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0558832469945854e-05,
+      "loss": 0.5937,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0520980486023623e-05,
+      "loss": 0.7637,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0483159395345454e-05,
+      "loss": 0.7735,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0445369212612254e-05,
+      "loss": 0.7064,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0407609952513007e-05,
+      "loss": 0.5882,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0369881629724586e-05,
+      "loss": 0.8045,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033218425891189e-05,
+      "loss": 0.6668,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0294517854727723e-05,
+      "loss": 0.7243,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025688243181295e-05,
+      "loss": 0.6471,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021927800479633e-05,
+      "loss": 0.6115,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0181704588294537e-05,
+      "loss": 0.6354,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.014416219691231e-05,
+      "loss": 0.6957,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0106650845242226e-05,
+      "loss": 0.7983,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0069170547864802e-05,
+      "loss": 0.8059,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0031721319348596e-05,
+      "loss": 0.8546,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994303174249884e-05,
+      "loss": 0.681,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9956916127113078e-05,
+      "loss": 0.6837,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9919560192470343e-05,
+      "loss": 0.6744,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9882235384841864e-05,
+      "loss": 0.6999,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9844941718735654e-05,
+      "loss": 0.7345,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9807679208647623e-05,
+      "loss": 0.7175,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9770447869061626e-05,
+      "loss": 0.6807,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9733247714449355e-05,
+      "loss": 0.7332,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9696078759270386e-05,
+      "loss": 0.6748,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9658941017972187e-05,
+      "loss": 0.6537,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9621834504990044e-05,
+      "loss": 0.7496,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9584759234747186e-05,
+      "loss": 0.5944,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9547715221654618e-05,
+      "loss": 0.6512,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9510702480111276e-05,
+      "loss": 0.7352,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9473721024503867e-05,
+      "loss": 0.5497,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9436770869206977e-05,
+      "loss": 0.6774,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9399852028582997e-05,
+      "loss": 0.6598,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936296451698214e-05,
+      "loss": 0.6372,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9326108348742543e-05,
+      "loss": 0.6875,
+      "step": 4181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.928928353819003e-05,
+      "loss": 0.6484,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9252490099638266e-05,
+      "loss": 0.5791,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9215728047388803e-05,
+      "loss": 0.782,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917899739573088e-05,
+      "loss": 0.682,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9142298158941642e-05,
+      "loss": 0.7345,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910563035128593e-05,
+      "loss": 0.7103,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9068993987016415e-05,
+      "loss": 0.7232,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9032389080373538e-05,
+      "loss": 0.694,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.899581564558548e-05,
+      "loss": 0.7759,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8959273696868262e-05,
+      "loss": 0.7135,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8922763248425578e-05,
+      "loss": 0.5814,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8886284314449e-05,
+      "loss": 0.8505,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8849836909117713e-05,
+      "loss": 0.6276,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8813421046598724e-05,
+      "loss": 0.5879,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8777036741046762e-05,
+      "loss": 0.6843,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874068400660427e-05,
+      "loss": 0.8505,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.870436285740149e-05,
+      "loss": 0.7568,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.866807330755631e-05,
+      "loss": 0.6515,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8631815371174345e-05,
+      "loss": 0.6265,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8595589062348982e-05,
+      "loss": 0.6795,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8559394395161223e-05,
+      "loss": 0.6569,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8523231383679885e-05,
+      "loss": 0.7027,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.848710004196138e-05,
+      "loss": 0.5889,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8451000384049844e-05,
+      "loss": 0.665,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8414932423977103e-05,
+      "loss": 0.7015,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837889617576264e-05,
+      "loss": 0.6007,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8342891653413673e-05,
+      "loss": 0.687,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8306918870925028e-05,
+      "loss": 0.8182,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827097784227917e-05,
+      "loss": 0.6911,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8235068581446346e-05,
+      "loss": 0.6495,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8199191102384317e-05,
+      "loss": 0.6776,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.816334541903856e-05,
+      "loss": 0.7633,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8127531545342146e-05,
+      "loss": 0.7419,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809174949521587e-05,
+      "loss": 0.6403,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8055999282568092e-05,
+      "loss": 0.7007,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.802028092129475e-05,
+      "loss": 0.6787,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7984594425279533e-05,
+      "loss": 0.7107,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.794893980839363e-05,
+      "loss": 0.6945,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7913317084495863e-05,
+      "loss": 0.7636,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7877726267432748e-05,
+      "loss": 0.6854,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784216737103821e-05,
+      "loss": 0.6968,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.780664040913397e-05,
+      "loss": 0.6997,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7771145395529187e-05,
+      "loss": 0.6964,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.773568234402072e-05,
+      "loss": 0.6631,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.770025126839291e-05,
+      "loss": 0.6486,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.766485218241768e-05,
+      "loss": 0.656,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7629485099854604e-05,
+      "loss": 0.6777,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.759415003445071e-05,
+      "loss": 0.7012,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7558846999940647e-05,
+      "loss": 0.842,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7523576010046584e-05,
+      "loss": 0.612,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7488337078478214e-05,
+      "loss": 0.655,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7453130218932856e-05,
+      "loss": 0.7671,
+      "step": 4233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7417955445095256e-05,
+      "loss": 0.8032,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.738281277063779e-05,
+      "loss": 0.7988,
+      "step": 4235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7347702209220264e-05,
+      "loss": 0.6905,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7312623774490054e-05,
+      "loss": 0.7046,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.727757748008204e-05,
+      "loss": 0.6519,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7242563339618577e-05,
+      "loss": 0.7564,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.72075813667096e-05,
+      "loss": 0.7434,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7172631574952446e-05,
+      "loss": 0.6847,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7137713977932026e-05,
+      "loss": 0.6612,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7102828589220687e-05,
+      "loss": 0.6835,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067975422378246e-05,
+      "loss": 0.6989,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7033154490952062e-05,
+      "loss": 0.711,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699836580847689e-05,
+      "loss": 0.6928,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6963609388474998e-05,
+      "loss": 0.6168,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6928885244456082e-05,
+      "loss": 0.6767,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.689419338991729e-05,
+      "loss": 0.757,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6859533838343287e-05,
+      "loss": 0.6935,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6824906603206082e-05,
+      "loss": 0.6867,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6790311697965212e-05,
+      "loss": 0.625,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.675574913606759e-05,
+      "loss": 0.6692,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672121893094758e-05,
+      "loss": 0.6532,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6686721096026958e-05,
+      "loss": 0.7616,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6652255644714886e-05,
+      "loss": 0.6551,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661782259040805e-05,
+      "loss": 0.7107,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.658342194649043e-05,
+      "loss": 0.5353,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.654905372633343e-05,
+      "loss": 0.6951,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.651471794329591e-05,
+      "loss": 0.7256,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6480414610724037e-05,
+      "loss": 0.5933,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.644614374195146e-05,
+      "loss": 0.5968,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6411905350299117e-05,
+      "loss": 0.7121,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.637769944907539e-05,
+      "loss": 0.6676,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6343526051575973e-05,
+      "loss": 0.678,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6309385171083968e-05,
+      "loss": 0.6345,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6275276820869845e-05,
+      "loss": 0.7471,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.62412010141914e-05,
+      "loss": 0.8007,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6207157764293778e-05,
+      "loss": 0.7569,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.617314708440951e-05,
+      "loss": 0.6436,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6139168987758434e-05,
+      "loss": 0.8086,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.610522348754773e-05,
+      "loss": 0.7124,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6071310596971857e-05,
+      "loss": 0.7512,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.603743032921272e-05,
+      "loss": 0.7417,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6003582697439446e-05,
+      "loss": 0.6291,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5969767714808482e-05,
+      "loss": 0.7191,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5935985394463637e-05,
+      "loss": 0.6726,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5902235749535976e-05,
+      "loss": 0.7109,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5868518793143862e-05,
+      "loss": 0.6521,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.583483453839303e-05,
+      "loss": 0.7613,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5801182998376364e-05,
+      "loss": 0.776,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5767564186174165e-05,
+      "loss": 0.7495,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.573397811485392e-05,
+      "loss": 0.5934,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.570042479747048e-05,
+      "loss": 0.6702,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5666904247065882e-05,
+      "loss": 0.6179,
+      "step": 4285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.563341647666945e-05,
+      "loss": 0.649,
+      "step": 4286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.55999614992978e-05,
+      "loss": 0.6295,
+      "step": 4287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5566539327954766e-05,
+      "loss": 0.7173,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5533149975631444e-05,
+      "loss": 0.7517,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5499793455306133e-05,
+      "loss": 0.7154,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.546646977994446e-05,
+      "loss": 0.6555,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5433178962499208e-05,
+      "loss": 0.7506,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.53999210159104e-05,
+      "loss": 0.7208,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5366695953105324e-05,
+      "loss": 0.7039,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.533350378699845e-05,
+      "loss": 0.6617,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.530034453049143e-05,
+      "loss": 0.6789,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5267218196473254e-05,
+      "loss": 0.6496,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5234124797819904e-05,
+      "loss": 0.6563,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5201064347394767e-05,
+      "loss": 0.7375,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5168036858048273e-05,
+      "loss": 0.6221,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5135042342618178e-05,
+      "loss": 0.7314,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5102080813929298e-05,
+      "loss": 0.6804,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5069152284793653e-05,
+      "loss": 0.6596,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5036256768010514e-05,
+      "loss": 0.721,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5003394276366245e-05,
+      "loss": 0.7569,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.497056482263438e-05,
+      "loss": 0.6685,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4937768419575626e-05,
+      "loss": 0.6482,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4905005079937829e-05,
+      "loss": 0.7324,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4872274816456034e-05,
+      "loss": 0.6576,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4839577641852343e-05,
+      "loss": 0.6617,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.48069135688361e-05,
+      "loss": 0.6962,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4774282610103696e-05,
+      "loss": 0.7105,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4741684778338671e-05,
+      "loss": 0.6969,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4709120086211726e-05,
+      "loss": 0.7273,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.467658854638062e-05,
+      "loss": 0.6595,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.46440901714903e-05,
+      "loss": 0.6406,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4611624974172755e-05,
+      "loss": 0.7755,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4579192967047085e-05,
+      "loss": 0.6512,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4546794162719569e-05,
+      "loss": 0.5664,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4514428573783457e-05,
+      "loss": 0.7211,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4482096212819207e-05,
+      "loss": 0.6745,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444979709239428e-05,
+      "loss": 0.6833,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4417531225063241e-05,
+      "loss": 0.7164,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4385298623367727e-05,
+      "loss": 0.6377,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4353099299836437e-05,
+      "loss": 0.8854,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4320933266985182e-05,
+      "loss": 0.7061,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4288800537316771e-05,
+      "loss": 0.7921,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4256701123321082e-05,
+      "loss": 0.7,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4224635037475097e-05,
+      "loss": 0.5981,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4192602292242785e-05,
+      "loss": 0.7452,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416060290007517e-05,
+      "loss": 0.6529,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4128636873410284e-05,
+      "loss": 0.697,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4096704224673285e-05,
+      "loss": 0.6937,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4064804966276245e-05,
+      "loss": 0.6256,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4032939110618292e-05,
+      "loss": 0.6728,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400110667008564e-05,
+      "loss": 0.7625,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3969307657051412e-05,
+      "loss": 0.6796,
+      "step": 4337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.393754208387581e-05,
+      "loss": 0.6894,
+      "step": 4338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3905809962906002e-05,
+      "loss": 0.7372,
+      "step": 4339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3874111306476168e-05,
+      "loss": 0.5918,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3842446126907471e-05,
+      "loss": 0.725,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3810814436508046e-05,
+      "loss": 0.6863,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3779216247573067e-05,
+      "loss": 0.6857,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3747651572384635e-05,
+      "loss": 0.6067,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.371612042321182e-05,
+      "loss": 0.7185,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3684622812310722e-05,
+      "loss": 0.7537,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3653158751924332e-05,
+      "loss": 0.6383,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.362172825428265e-05,
+      "loss": 0.7932,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3590331331602568e-05,
+      "loss": 0.7015,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3558967996088035e-05,
+      "loss": 0.6101,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3527638259929842e-05,
+      "loss": 0.7835,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3496342135305751e-05,
+      "loss": 0.6247,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3465079634380494e-05,
+      "loss": 0.6932,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.34338507693057e-05,
+      "loss": 0.5923,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3402655552219901e-05,
+      "loss": 0.7431,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.337149399524864e-05,
+      "loss": 0.7332,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3340366110504242e-05,
+      "loss": 0.6264,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3309271910086074e-05,
+      "loss": 0.7387,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.327821140608031e-05,
+      "loss": 0.6769,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3247184610560116e-05,
+      "loss": 0.6482,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3216191535585488e-05,
+      "loss": 0.6653,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3185232193203312e-05,
+      "loss": 0.6896,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3154306595447442e-05,
+      "loss": 0.5776,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3123414754338514e-05,
+      "loss": 0.6854,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3092556681884105e-05,
+      "loss": 0.7516,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3061732390078652e-05,
+      "loss": 0.7689,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3030941890903436e-05,
+      "loss": 0.6353,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3000185196326664e-05,
+      "loss": 0.703,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2969462318303338e-05,
+      "loss": 0.6214,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2938773268775372e-05,
+      "loss": 0.6658,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.290811805967148e-05,
+      "loss": 0.6643,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2877496702907232e-05,
+      "loss": 0.6486,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2846909210385117e-05,
+      "loss": 0.7595,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281635559399431e-05,
+      "loss": 0.6713,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2785835865610973e-05,
+      "loss": 0.5588,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2755350037097991e-05,
+      "loss": 0.5998,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2724898120305106e-05,
+      "loss": 0.7979,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.269448012706892e-05,
+      "loss": 0.6568,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2664096069212762e-05,
+      "loss": 0.6781,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2633745958546861e-05,
+      "loss": 0.7138,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2603429806868205e-05,
+      "loss": 0.5769,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2573147625960569e-05,
+      "loss": 0.6478,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2542899427594546e-05,
+      "loss": 0.7316,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2512685223527487e-05,
+      "loss": 0.6121,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.248250502550361e-05,
+      "loss": 0.6281,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.245235884525383e-05,
+      "loss": 0.6898,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2422246694495898e-05,
+      "loss": 0.7775,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2392168584934283e-05,
+      "loss": 0.7076,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2362124528260277e-05,
+      "loss": 0.6572,
+      "step": 4389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2332114536151894e-05,
+      "loss": 0.6383,
+      "step": 4390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2302138620273907e-05,
+      "loss": 0.801,
+      "step": 4391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.227219679227789e-05,
+      "loss": 0.6587,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2242289063802136e-05,
+      "loss": 0.7023,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2212415446471647e-05,
+      "loss": 0.6886,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2182575951898256e-05,
+      "loss": 0.6904,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.215277059168044e-05,
+      "loss": 0.6563,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2122999377403476e-05,
+      "loss": 0.6756,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.209326232063933e-05,
+      "loss": 0.7621,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2063559432946714e-05,
+      "loss": 0.7488,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2033890725871023e-05,
+      "loss": 0.6721,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2004256210944376e-05,
+      "loss": 0.6493,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1974655899685672e-05,
+      "loss": 0.6919,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1945089803600418e-05,
+      "loss": 0.6572,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1915557934180854e-05,
+      "loss": 0.6431,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1886060302905955e-05,
+      "loss": 0.6451,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1856596921241358e-05,
+      "loss": 0.6916,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1827167800639361e-05,
+      "loss": 0.6262,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1797772952538965e-05,
+      "loss": 0.6874,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176841238836589e-05,
+      "loss": 0.6704,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1739086119532472e-05,
+      "loss": 0.6683,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1709794157437737e-05,
+      "loss": 0.5687,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1680536513467388e-05,
+      "loss": 0.7729,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1651313198993784e-05,
+      "loss": 0.7505,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1622124225375908e-05,
+      "loss": 0.6919,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159296960395947e-05,
+      "loss": 0.5982,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1563849346076717e-05,
+      "loss": 0.7112,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.153476346304666e-05,
+      "loss": 0.7073,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1505711966174837e-05,
+      "loss": 0.6684,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1476694866753512e-05,
+      "loss": 0.6429,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1447712176061543e-05,
+      "loss": 0.6782,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1418763905364361e-05,
+      "loss": 0.6761,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1389850065914131e-05,
+      "loss": 0.6679,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1360970668949533e-05,
+      "loss": 0.779,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.133212572569592e-05,
+      "loss": 0.6466,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130331524736521e-05,
+      "loss": 0.7062,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1274539245155925e-05,
+      "loss": 0.7466,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124579773025326e-05,
+      "loss": 0.7774,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1217090713828904e-05,
+      "loss": 0.6629,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1188418207041217e-05,
+      "loss": 0.7366,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1159780221035109e-05,
+      "loss": 0.605,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1131176766942042e-05,
+      "loss": 0.7225,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1102607855880154e-05,
+      "loss": 0.682,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1074073498953996e-05,
+      "loss": 0.5931,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1045573707254863e-05,
+      "loss": 0.6374,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1017108491860484e-05,
+      "loss": 0.6999,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0988677863835239e-05,
+      "loss": 0.7364,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0960281834230002e-05,
+      "loss": 0.6294,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0931920414082198e-05,
+      "loss": 0.7555,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0903593614415875e-05,
+      "loss": 0.7691,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0875301446241538e-05,
+      "loss": 0.7697,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0847043920556265e-05,
+      "loss": 0.7116,
+      "step": 4441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0818821048343663e-05,
+      "loss": 0.6602,
+      "step": 4442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0790632840573866e-05,
+      "loss": 0.6965,
+      "step": 4443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0762479308203566e-05,
+      "loss": 0.727,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073436046217593e-05,
+      "loss": 0.6477,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0706276313420694e-05,
+      "loss": 0.6361,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0678226872854057e-05,
+      "loss": 0.8242,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0650212151378736e-05,
+      "loss": 0.6915,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0622232159884005e-05,
+      "loss": 0.6859,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0594286909245543e-05,
+      "loss": 0.6802,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.056637641032563e-05,
+      "loss": 0.6447,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0538500673972973e-05,
+      "loss": 0.6648,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0510659711022753e-05,
+      "loss": 0.6236,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0482853532296711e-05,
+      "loss": 0.6594,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0455082148602979e-05,
+      "loss": 0.7956,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0427345570736247e-05,
+      "loss": 0.7412,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0399643809477621e-05,
+      "loss": 0.6839,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0371976875594669e-05,
+      "loss": 0.6091,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0344344779841475e-05,
+      "loss": 0.6983,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0316747532958493e-05,
+      "loss": 0.7056,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0289185145672742e-05,
+      "loss": 0.6258,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026165762869762e-05,
+      "loss": 0.7024,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0234164992732952e-05,
+      "loss": 0.6819,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0206707248465086e-05,
+      "loss": 0.6278,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0179284406566737e-05,
+      "loss": 0.754,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0151896477697088e-05,
+      "loss": 0.678,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0124543472501712e-05,
+      "loss": 0.7611,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0097225401612675e-05,
+      "loss": 0.6541,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0069942275648404e-05,
+      "loss": 0.6928,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0042694105213745e-05,
+      "loss": 0.5612,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.001548090090002e-05,
+      "loss": 0.7512,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.988302673284878e-06,
+      "loss": 0.7041,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.961159432932442e-06,
+      "loss": 0.7549,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.9340511903932e-06,
+      "loss": 0.7235,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.906977956204e-06,
+      "loss": 0.6917,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.879939740888155e-06,
+      "loss": 0.6981,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.852936554955328e-06,
+      "loss": 0.6268,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.825968408901575e-06,
+      "loss": 0.71,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.799035313209338e-06,
+      "loss": 0.6746,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.772137278347394e-06,
+      "loss": 0.7179,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.745274314770969e-06,
+      "loss": 0.8126,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.718446432921591e-06,
+      "loss": 0.6833,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.691653643227161e-06,
+      "loss": 0.5918,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.664895956101949e-06,
+      "loss": 0.7168,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.638173381946624e-06,
+      "loss": 0.7863,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.611485931148123e-06,
+      "loss": 0.5977,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58483361407978e-06,
+      "loss": 0.6631,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558216441101298e-06,
+      "loss": 0.6591,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.531634422558656e-06,
+      "loss": 0.6749,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.505087568784199e-06,
+      "loss": 0.7712,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.478575890096664e-06,
+      "loss": 0.6285,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.452099396800962e-06,
+      "loss": 0.7205,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.425658099188483e-06,
+      "loss": 0.6586,
+      "step": 4493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.399252007536862e-06,
+      "loss": 0.7562,
+      "step": 4494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.372881132110068e-06,
+      "loss": 0.6804,
+      "step": 4495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.346545483158387e-06,
+      "loss": 0.6425,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.320245070918365e-06,
+      "loss": 0.7583,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.293979905612936e-06,
+      "loss": 0.6956,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.267749997451259e-06,
+      "loss": 0.6446,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.241555356628828e-06,
+      "loss": 0.6821,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.215395993327414e-06,
+      "loss": 0.6513,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.189271917715047e-06,
+      "loss": 0.6833,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.16318313994613e-06,
+      "loss": 0.692,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.137129670161238e-06,
+      "loss": 0.6241,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.111111518487314e-06,
+      "loss": 0.536,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.085128695037515e-06,
+      "loss": 0.5829,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.059181209911272e-06,
+      "loss": 0.6743,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.033269073194316e-06,
+      "loss": 0.6313,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.007392294958573e-06,
+      "loss": 0.636,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.981550885262303e-06,
+      "loss": 0.7247,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.95574485414996e-06,
+      "loss": 0.6513,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.92997421165227e-06,
+      "loss": 0.6849,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.904238967786206e-06,
+      "loss": 0.6494,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.878539132554952e-06,
+      "loss": 0.5788,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.852874715948e-06,
+      "loss": 0.7288,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827245727941003e-06,
+      "loss": 0.7013,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.801652178495856e-06,
+      "loss": 0.6477,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.776094077560704e-06,
+      "loss": 0.7925,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750571435069887e-06,
+      "loss": 0.6943,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.725084260943993e-06,
+      "loss": 0.6063,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.699632565089811e-06,
+      "loss": 0.6901,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.674216357400288e-06,
+      "loss": 0.6251,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.648835647754694e-06,
+      "loss": 0.5903,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623490446018379e-06,
+      "loss": 0.5692,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.598180762042973e-06,
+      "loss": 0.6055,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.57290660566623e-06,
+      "loss": 0.7192,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.547667986712183e-06,
+      "loss": 0.7113,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.522464914990968e-06,
+      "loss": 0.7444,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.497297400298954e-06,
+      "loss": 0.662,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.472165452418679e-06,
+      "loss": 0.6348,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.447069081118842e-06,
+      "loss": 0.6679,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.422008296154326e-06,
+      "loss": 0.7006,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.396983107266199e-06,
+      "loss": 0.6824,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.371993524181643e-06,
+      "loss": 0.6928,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.347039556614045e-06,
+      "loss": 0.6439,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.322121214262924e-06,
+      "loss": 0.6407,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.297238506813964e-06,
+      "loss": 0.7915,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.27239144393902e-06,
+      "loss": 0.624,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.24758003529601e-06,
+      "loss": 0.646,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.222804290529107e-06,
+      "loss": 0.6878,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198064219268553e-06,
+      "loss": 0.6278,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.173359831130712e-06,
+      "loss": 0.722,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.148691135718111e-06,
+      "loss": 0.6414,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124058142619405e-06,
+      "loss": 0.6257,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.099460861409358e-06,
+      "loss": 0.7254,
+      "step": 4545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.074899301648841e-06,
+      "loss": 0.6967,
+      "step": 4546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.050373472884875e-06,
+      "loss": 0.8034,
+      "step": 4547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.025883384650567e-06,
+      "loss": 0.6716,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.001429046465104e-06,
+      "loss": 0.5951,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.97701046783389e-06,
+      "loss": 0.6597,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.952627658248258e-06,
+      "loss": 0.6932,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.92828062718578e-06,
+      "loss": 0.6859,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.903969384110044e-06,
+      "loss": 0.8029,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.879693938470788e-06,
+      "loss": 0.642,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.855454299703791e-06,
+      "loss": 0.7033,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.831250477230889e-06,
+      "loss": 0.5945,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.807082480460082e-06,
+      "loss": 0.6387,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.782950318785364e-06,
+      "loss": 0.7069,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.758854001586835e-06,
+      "loss": 0.6461,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.734793538230656e-06,
+      "loss": 0.7382,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.710768938069046e-06,
+      "loss": 0.7066,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.686780210440314e-06,
+      "loss": 0.5837,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.662827364668768e-06,
+      "loss": 0.6899,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.638910410064848e-06,
+      "loss": 0.7031,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.615029355924974e-06,
+      "loss": 0.7518,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5911842115316214e-06,
+      "loss": 0.6649,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.567374986153375e-06,
+      "loss": 0.6804,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.543601689044732e-06,
+      "loss": 0.5889,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.519864329446358e-06,
+      "loss": 0.6763,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.496162916584859e-06,
+      "loss": 0.6108,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4724974596729e-06,
+      "loss": 0.7096,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.44886796790919e-06,
+      "loss": 0.7047,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4252744504784145e-06,
+      "loss": 0.6351,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.401716916551316e-06,
+      "loss": 0.7311,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.378195375284636e-06,
+      "loss": 0.6125,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3547098358211166e-06,
+      "loss": 0.7152,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3312603072895e-06,
+      "loss": 0.6931,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.307846798804552e-06,
+      "loss": 0.7229,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284469319467036e-06,
+      "loss": 0.648,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2611278783636934e-06,
+      "loss": 0.6715,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.237822484567291e-06,
+      "loss": 0.6915,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.214553147136549e-06,
+      "loss": 0.6815,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.191319875116165e-06,
+      "loss": 0.7538,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.168122677536893e-06,
+      "loss": 0.6681,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.144961563415342e-06,
+      "loss": 0.6097,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.121836541754212e-06,
+      "loss": 0.6318,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.098747621542101e-06,
+      "loss": 0.7584,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.075694811753608e-06,
+      "loss": 0.6504,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.052678121349299e-06,
+      "loss": 0.7107,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.029697559275661e-06,
+      "loss": 0.6165,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.006753134465194e-06,
+      "loss": 0.5837,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.983844855836319e-06,
+      "loss": 0.6132,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.960972732293414e-06,
+      "loss": 0.6733,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.938136772726778e-06,
+      "loss": 0.6028,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.915336986012677e-06,
+      "loss": 0.6125,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.892573381013357e-06,
+      "loss": 0.6303,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.869845966576927e-06,
+      "loss": 0.6851,
+      "step": 4597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.847154751537443e-06,
+      "loss": 0.6243,
+      "step": 4598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.824499744714952e-06,
+      "loss": 0.6672,
+      "step": 4599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.801880954915374e-06,
+      "loss": 0.8312,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.779298390930544e-06,
+      "loss": 0.7276,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.756752061538229e-06,
+      "loss": 0.7124,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7342419755021425e-06,
+      "loss": 0.6601,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.711768141571862e-06,
+      "loss": 0.7227,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.689330568482888e-06,
+      "loss": 0.6792,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.666929264956679e-06,
+      "loss": 0.6985,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.644564239700513e-06,
+      "loss": 0.6909,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.622235501407592e-06,
+      "loss": 0.6707,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.599943058757085e-06,
+      "loss": 0.6534,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.577686920413928e-06,
+      "loss": 0.6854,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.555467095029044e-06,
+      "loss": 0.7172,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.533283591239203e-06,
+      "loss": 0.6198,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5111364176670745e-06,
+      "loss": 0.6449,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.489025582921182e-06,
+      "loss": 0.7412,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.46695109559593e-06,
+      "loss": 0.6754,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.444912964271632e-06,
+      "loss": 0.8053,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.422911197514414e-06,
+      "loss": 0.6086,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4009458038763016e-06,
+      "loss": 0.5948,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.379016791895176e-06,
+      "loss": 0.713,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.357124170094752e-06,
+      "loss": 0.6547,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.335267946984658e-06,
+      "loss": 0.6133,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.313448131060307e-06,
+      "loss": 0.6663,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.291664730803026e-06,
+      "loss": 0.7085,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2699177546799435e-06,
+      "loss": 0.6965,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.248207211144008e-06,
+      "loss": 0.6985,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.226533108634103e-06,
+      "loss": 0.6823,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204895455574811e-06,
+      "loss": 0.688,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.183294260376671e-06,
+      "loss": 0.6234,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1617295314359665e-06,
+      "loss": 0.6922,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.140201277134872e-06,
+      "loss": 0.6506,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.118709505841324e-06,
+      "loss": 0.6281,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097254225909122e-06,
+      "loss": 0.7925,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.075835445677869e-06,
+      "loss": 0.7452,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0544531734729625e-06,
+      "loss": 0.5561,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.033107417605632e-06,
+      "loss": 0.6549,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.011798186372919e-06,
+      "loss": 0.7155,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.990525488057608e-06,
+      "loss": 0.7532,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.969289330928385e-06,
+      "loss": 0.6137,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.948089723239636e-06,
+      "loss": 0.5572,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.926926673231625e-06,
+      "loss": 0.6412,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.905800189130328e-06,
+      "loss": 0.7555,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.884710279147542e-06,
+      "loss": 0.6354,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.863656951480878e-06,
+      "loss": 0.5671,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.842640214313666e-06,
+      "loss": 0.6311,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821660075815061e-06,
+      "loss": 0.6486,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.800716544139972e-06,
+      "loss": 0.7107,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.779809627429078e-06,
+      "loss": 0.768,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.758939333808855e-06,
+      "loss": 0.6899,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.73810567139148e-06,
+      "loss": 0.6633,
+      "step": 4649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.717308648274977e-06,
+      "loss": 0.654,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.696548272543068e-06,
+      "loss": 0.6233,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.67582455226523e-06,
+      "loss": 0.6022,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.655137495496721e-06,
+      "loss": 0.6579,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.634487110278519e-06,
+      "loss": 0.701,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.613873404637382e-06,
+      "loss": 0.728,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5932963865858e-06,
+      "loss": 0.8495,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.572756064121953e-06,
+      "loss": 0.7325,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5522524452298555e-06,
+      "loss": 0.7405,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.531785537879142e-06,
+      "loss": 0.6891,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.511355350025305e-06,
+      "loss": 0.5916,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.490961889609425e-06,
+      "loss": 0.62,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.470605164558407e-06,
+      "loss": 0.6956,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.450285182784853e-06,
+      "loss": 0.7081,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.430001952187047e-06,
+      "loss": 0.6667,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.409755480649059e-06,
+      "loss": 0.719,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389545776040583e-06,
+      "loss": 0.6236,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.369372846217102e-06,
+      "loss": 0.7278,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.349236699019766e-06,
+      "loss": 0.6619,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.329137342275403e-06,
+      "loss": 0.5952,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3090747837966085e-06,
+      "loss": 0.8239,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.289049031381599e-06,
+      "loss": 0.7442,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.269060092814348e-06,
+      "loss": 0.6324,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.249107975864487e-06,
+      "loss": 0.7343,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.229192688287332e-06,
+      "loss": 0.5851,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.209314237823903e-06,
+      "loss": 0.7221,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.189472632200898e-06,
+      "loss": 0.6751,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.169667879130657e-06,
+      "loss": 0.6748,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.149899986311247e-06,
+      "loss": 0.6393,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1301689614263845e-06,
+      "loss": 0.7872,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.11047481214546e-06,
+      "loss": 0.676,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.090817546123505e-06,
+      "loss": 0.6385,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.071197171001263e-06,
+      "loss": 0.5854,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.05161369440511e-06,
+      "loss": 0.7171,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.032067123947049e-06,
+      "loss": 0.6756,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.012557467224832e-06,
+      "loss": 0.619,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.993084731821729e-06,
+      "loss": 0.7576,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.973648925306773e-06,
+      "loss": 0.6451,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.954250055234588e-06,
+      "loss": 0.6561,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.934888129145477e-06,
+      "loss": 0.7254,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.915563154565328e-06,
+      "loss": 0.6358,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896275139005702e-06,
+      "loss": 0.6203,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.877024089963822e-06,
+      "loss": 0.6856,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.85781001492247e-06,
+      "loss": 0.5583,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.838632921350128e-06,
+      "loss": 0.6028,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.819492816700844e-06,
+      "loss": 0.6459,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800389708414321e-06,
+      "loss": 0.6663,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.781323603915899e-06,
+      "loss": 0.5841,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.762294510616483e-06,
+      "loss": 0.6302,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7433024359126446e-06,
+      "loss": 0.6599,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.724347387186545e-06,
+      "loss": 0.7105,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.705429371805925e-06,
+      "loss": 0.6829,
+      "step": 4701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.686548397124202e-06,
+      "loss": 0.579,
+      "step": 4702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6677044704802945e-06,
+      "loss": 0.7164,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.648897599198821e-06,
+      "loss": 0.6681,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.630127790589944e-06,
+      "loss": 0.6601,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.611395051949407e-06,
+      "loss": 0.7286,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.592699390558597e-06,
+      "loss": 0.7128,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5740408136844326e-06,
+      "loss": 0.8439,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555419328579469e-06,
+      "loss": 0.6546,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5368349424818044e-06,
+      "loss": 0.6269,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.518287662615128e-06,
+      "loss": 0.7946,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4997774961887175e-06,
+      "loss": 0.8072,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4813044503973944e-06,
+      "loss": 0.6632,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.462868532421616e-06,
+      "loss": 0.6836,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.444469749427316e-06,
+      "loss": 0.7243,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.426108108566085e-06,
+      "loss": 0.6869,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.407783616975014e-06,
+      "loss": 0.7492,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.389496281776762e-06,
+      "loss": 0.631,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.371246110079608e-06,
+      "loss": 0.7583,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.353033108977278e-06,
+      "loss": 0.7671,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.334857285549143e-06,
+      "loss": 0.6402,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.316718646860074e-06,
+      "loss": 0.6139,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298617199960508e-06,
+      "loss": 0.6441,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2805529518864185e-06,
+      "loss": 0.757,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.262525909659309e-06,
+      "loss": 0.6013,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.244536080286265e-06,
+      "loss": 0.7301,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2265834707598596e-06,
+      "loss": 0.8069,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.208668088058221e-06,
+      "loss": 0.638,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.190789939144979e-06,
+      "loss": 0.6882,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.172949030969331e-06,
+      "loss": 0.881,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1551453704659845e-06,
+      "loss": 0.7058,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1373789645551716e-06,
+      "loss": 0.6414,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.119649820142613e-06,
+      "loss": 0.5973,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.101957944119594e-06,
+      "loss": 0.6761,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.084303343362894e-06,
+      "loss": 0.6513,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.066686024734778e-06,
+      "loss": 0.6677,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.049105995083058e-06,
+      "loss": 0.7348,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0315632612410335e-06,
+      "loss": 0.6589,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0140578300275175e-06,
+      "loss": 0.6378,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.996589708246789e-06,
+      "loss": 0.5872,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9791589026886845e-06,
+      "loss": 0.7686,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9617654201284935e-06,
+      "loss": 0.6582,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9444092673269875e-06,
+      "loss": 0.6552,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.927090451030491e-06,
+      "loss": 0.7094,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909808977970719e-06,
+      "loss": 0.6731,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.892564854864966e-06,
+      "loss": 0.7091,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.875358088415937e-06,
+      "loss": 0.6952,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.858188685311892e-06,
+      "loss": 0.6861,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8410566522264825e-06,
+      "loss": 0.7101,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8239619958188924e-06,
+      "loss": 0.7015,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8069047227337617e-06,
+      "loss": 0.7992,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7898848396012097e-06,
+      "loss": 0.7167,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.772902353036789e-06,
+      "loss": 0.7879,
+      "step": 4753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.755957269641541e-06,
+      "loss": 0.6828,
+      "step": 4754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7390495960019643e-06,
+      "loss": 0.6355,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.722179338690046e-06,
+      "loss": 0.7722,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.705346504263163e-06,
+      "loss": 0.7758,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.688551099264226e-06,
+      "loss": 0.7228,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6717931302215237e-06,
+      "loss": 0.6558,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6550726036488236e-06,
+      "loss": 0.6033,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.638389526045383e-06,
+      "loss": 0.6893,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6217439038958133e-06,
+      "loss": 0.7645,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6051357436702404e-06,
+      "loss": 0.6012,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5885650518241887e-06,
+      "loss": 0.6304,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5720318347986503e-06,
+      "loss": 0.5937,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5555360990200292e-06,
+      "loss": 0.7597,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.539077850900152e-06,
+      "loss": 0.6349,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.522657096836324e-06,
+      "loss": 0.6278,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5062738432112054e-06,
+      "loss": 0.6575,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4899280963929247e-06,
+      "loss": 0.6678,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4736198627350225e-06,
+      "loss": 0.7849,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.457349148576461e-06,
+      "loss": 0.6129,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4411159602416143e-06,
+      "loss": 0.7162,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4249203040402578e-06,
+      "loss": 0.6349,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4087621862676113e-06,
+      "loss": 0.6575,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.392641613204295e-06,
+      "loss": 0.6039,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3765585911162746e-06,
+      "loss": 0.6725,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.360513126255038e-06,
+      "loss": 0.6979,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3445052248573526e-06,
+      "loss": 0.7623,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.328534893145463e-06,
+      "loss": 0.7228,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3126021373269813e-06,
+      "loss": 0.6349,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.296706963594909e-06,
+      "loss": 0.7727,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2808493781276705e-06,
+      "loss": 0.6279,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.265029387089047e-06,
+      "loss": 0.7389,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2492469966282303e-06,
+      "loss": 0.8086,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2335022128797687e-06,
+      "loss": 0.7354,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.217795041963623e-06,
+      "loss": 0.6077,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2021254899851195e-06,
+      "loss": 0.691,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1864935630349426e-06,
+      "loss": 0.7152,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.170899267189209e-06,
+      "loss": 0.617,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1553426085093485e-06,
+      "loss": 0.6735,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1398235930421906e-06,
+      "loss": 0.7072,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1243422268199317e-06,
+      "loss": 0.715,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1088985158601147e-06,
+      "loss": 0.7632,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.093492466165693e-06,
+      "loss": 0.6978,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.078124083724909e-06,
+      "loss": 0.6836,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.06279337451143e-06,
+      "loss": 0.6945,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0475003444842555e-06,
+      "loss": 0.7008,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0322449995877187e-06,
+      "loss": 0.704,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.017027345751555e-06,
+      "loss": 0.6881,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0018473888907993e-06,
+      "loss": 0.7623,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9867051349058318e-06,
+      "loss": 0.6065,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.971600589682455e-06,
+      "loss": 0.5206,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9565337590916953e-06,
+      "loss": 0.6554,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.941504648990023e-06,
+      "loss": 0.71,
+      "step": 4805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9265132652191774e-06,
+      "loss": 0.7593,
+      "step": 4806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.911559613606296e-06,
+      "loss": 0.7597,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.896643699963797e-06,
+      "loss": 0.6917,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.881765530089442e-06,
+      "loss": 0.6267,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8669251097663386e-06,
+      "loss": 0.8485,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.852122444762917e-06,
+      "loss": 0.645,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.837357540832919e-06,
+      "loss": 0.6756,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8226304037154207e-06,
+      "loss": 0.6357,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8079410391347872e-06,
+      "loss": 0.7089,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7932894528007624e-06,
+      "loss": 0.6956,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7786756504083577e-06,
+      "loss": 0.6968,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.764099637637918e-06,
+      "loss": 0.6626,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.749561420155089e-06,
+      "loss": 0.6802,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.735061003610828e-06,
+      "loss": 0.6269,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.720598393641416e-06,
+      "loss": 0.5987,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.706173595868389e-06,
+      "loss": 0.6677,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6917866158986613e-06,
+      "loss": 0.6414,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.677437459324372e-06,
+      "loss": 0.7117,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.663126131723015e-06,
+      "loss": 0.6836,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.648852638657362e-06,
+      "loss": 0.6233,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.634616985675453e-06,
+      "loss": 0.6364,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6204191783106513e-06,
+      "loss": 0.7456,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6062592220816197e-06,
+      "loss": 0.6407,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5921371224922552e-06,
+      "loss": 0.7278,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5780528850317896e-06,
+      "loss": 0.6542,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5640065151747217e-06,
+      "loss": 0.6491,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.549998018380828e-06,
+      "loss": 0.6469,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5360274000951644e-06,
+      "loss": 0.8107,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.522094665748087e-06,
+      "loss": 0.6244,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.508199820755186e-06,
+      "loss": 0.6803,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49434287051733e-06,
+      "loss": 0.6362,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4805238204207105e-06,
+      "loss": 0.6273,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4667426758367194e-06,
+      "loss": 0.633,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.45299944212205e-06,
+      "loss": 0.6208,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4392941246186516e-06,
+      "loss": 0.6397,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4256267286537405e-06,
+      "loss": 0.6458,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4119972595398e-06,
+      "loss": 0.74,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3984057225745484e-06,
+      "loss": 0.6756,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3848521230409815e-06,
+      "loss": 0.677,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3713364662073414e-06,
+      "loss": 0.5341,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3578587573271137e-06,
+      "loss": 0.7636,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.344419001639053e-06,
+      "loss": 0.7609,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.331017204367125e-06,
+      "loss": 0.7101,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3176533707205962e-06,
+      "loss": 0.7701,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.304327505893933e-06,
+      "loss": 0.6569,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291039615066848e-06,
+      "loss": 0.5484,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2777897034043206e-06,
+      "loss": 0.7354,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264577776056531e-06,
+      "loss": 0.7541,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2514038381589384e-06,
+      "loss": 0.6953,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23826789483218e-06,
+      "loss": 0.6895,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2251699511821822e-06,
+      "loss": 0.7275,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2121100123000727e-06,
+      "loss": 0.6002,
+      "step": 4857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1990880832621795e-06,
+      "loss": 0.8014,
+      "step": 4858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1861041691301097e-06,
+      "loss": 0.6171,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1731582749506707e-06,
+      "loss": 0.7114,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1602504057558815e-06,
+      "loss": 0.6845,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1473805665630063e-06,
+      "loss": 0.7497,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1345487623744886e-06,
+      "loss": 0.6947,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.121754998178027e-06,
+      "loss": 0.7389,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1089992789464884e-06,
+      "loss": 0.6737,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.096281609638029e-06,
+      "loss": 0.7228,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.083601995195927e-06,
+      "loss": 0.5686,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.070960440548719e-06,
+      "loss": 0.6788,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0583569506101517e-06,
+      "loss": 0.8239,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0457915302791397e-06,
+      "loss": 0.5804,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.033264184439854e-06,
+      "loss": 0.5776,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0207749179615875e-06,
+      "loss": 0.7877,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0083237356989247e-06,
+      "loss": 0.6546,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.995910642491572e-06,
+      "loss": 0.7205,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.983535643164458e-06,
+      "loss": 0.8046,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9711987425277356e-06,
+      "loss": 0.6146,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.958899945376691e-06,
+      "loss": 0.6548,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9466392564918333e-06,
+      "loss": 0.6592,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.934416680638873e-06,
+      "loss": 0.7071,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.922232222568654e-06,
+      "loss": 0.6618,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9100858870172544e-06,
+      "loss": 0.6004,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.89797767870592e-06,
+      "loss": 0.7624,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8859076023410638e-06,
+      "loss": 0.6859,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8738756626142994e-06,
+      "loss": 0.6045,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.861881864202386e-06,
+      "loss": 0.6918,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8499262117672945e-06,
+      "loss": 0.6458,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8380087099561295e-06,
+      "loss": 0.7635,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8261293634012078e-06,
+      "loss": 0.7025,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8142881767199804e-06,
+      "loss": 0.6703,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8024851545150768e-06,
+      "loss": 0.6374,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7907203013742934e-06,
+      "loss": 0.7822,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.778993621870606e-06,
+      "loss": 0.7316,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7673051205621238e-06,
+      "loss": 0.6482,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.755654801992146e-06,
+      "loss": 0.6689,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7440426706890833e-06,
+      "loss": 0.7227,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.732468731166581e-06,
+      "loss": 0.6502,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7209329879233515e-06,
+      "loss": 0.6525,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7094354454433303e-06,
+      "loss": 0.7444,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6979761081955537e-06,
+      "loss": 0.6771,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6865549806342363e-06,
+      "loss": 0.756,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6751720671987381e-06,
+      "loss": 0.7543,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6638273723135644e-06,
+      "loss": 0.7114,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6525209003883657e-06,
+      "loss": 0.7111,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6412526558179153e-06,
+      "loss": 0.6679,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6300226429821652e-06,
+      "loss": 0.6215,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6188308662461681e-06,
+      "loss": 0.6623,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076773299601333e-06,
+      "loss": 0.6558,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5965620384594148e-06,
+      "loss": 0.7266,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.585484996064479e-06,
+      "loss": 0.6326,
+      "step": 4909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5744462070809595e-06,
+      "loss": 0.7012,
+      "step": 4910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5634456757995908e-06,
+      "loss": 0.7107,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5524834064962191e-06,
+      "loss": 0.6915,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5415594034318803e-06,
+      "loss": 0.6649,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.530673670852678e-06,
+      "loss": 0.637,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.519826212989861e-06,
+      "loss": 0.6608,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.509017034059823e-06,
+      "loss": 0.668,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4982461382640257e-06,
+      "loss": 0.6844,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4875135297891086e-06,
+      "loss": 0.698,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4768192128067904e-06,
+      "loss": 0.7141,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4661631914739239e-06,
+      "loss": 0.6193,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4555454699324733e-06,
+      "loss": 0.64,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4449660523095154e-06,
+      "loss": 0.7071,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4344249427172163e-06,
+      "loss": 0.8073,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4239221452528872e-06,
+      "loss": 0.6876,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4134576639989294e-06,
+      "loss": 0.7353,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.403031503022867e-06,
+      "loss": 0.6517,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3926436663772914e-06,
+      "loss": 0.8272,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3822941580999394e-06,
+      "loss": 0.6985,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3719829822136266e-06,
+      "loss": 0.5934,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3617101427262913e-06,
+      "loss": 0.7321,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3514756436309173e-06,
+      "loss": 0.6293,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3412794889056446e-06,
+      "loss": 0.5954,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3311216825136917e-06,
+      "loss": 0.6336,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3210022284033562e-06,
+      "loss": 0.6167,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.310921130508036e-06,
+      "loss": 0.7756,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.300878392746241e-06,
+      "loss": 0.8166,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2908740190215152e-06,
+      "loss": 0.7165,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2809080132225704e-06,
+      "loss": 0.8321,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2709803792231301e-06,
+      "loss": 0.7081,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2610911208820409e-06,
+      "loss": 0.6506,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.251240242043239e-06,
+      "loss": 0.6364,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2414277465357172e-06,
+      "loss": 0.6901,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2316536381735576e-06,
+      "loss": 0.6967,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2219179207559328e-06,
+      "loss": 0.7281,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2122205980671043e-06,
+      "loss": 0.7364,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2025616738763567e-06,
+      "loss": 0.6561,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1929411519380984e-06,
+      "loss": 0.6826,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1833590359918046e-06,
+      "loss": 0.6853,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1738153297619958e-06,
+      "loss": 0.8338,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1643100369582937e-06,
+      "loss": 0.6582,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1548431612753652e-06,
+      "loss": 0.6636,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1454147063929776e-06,
+      "loss": 0.571,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1360246759759108e-06,
+      "loss": 0.6199,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1266730736740561e-06,
+      "loss": 0.7484,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1173599031223613e-06,
+      "loss": 0.6663,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.108085167940809e-06,
+      "loss": 0.6957,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0988488717344592e-06,
+      "loss": 0.681,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0896510180934406e-06,
+      "loss": 0.7605,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0804916105929374e-06,
+      "loss": 0.6734,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.071370652793169e-06,
+      "loss": 0.6224,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0622881482394098e-06,
+      "loss": 0.6993,
+      "step": 4961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0532441004620364e-06,
+      "loss": 0.6765,
+      "step": 4962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0442385129764143e-06,
+      "loss": 0.739,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0352713892829991e-06,
+      "loss": 0.7817,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0263427328672692e-06,
+      "loss": 0.5881,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.017452547199782e-06,
+      "loss": 0.7283,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0086008357361177e-06,
+      "loss": 0.5976,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.997876019169017e-07,
+      "loss": 0.7165,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.910128491678271e-07,
+      "loss": 0.5642,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.822765808995992e-07,
+      "loss": 0.6622,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.735788005079793e-07,
+      "loss": 0.706,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.64919511373774e-07,
+      "loss": 0.7282,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.562987168628134e-07,
+      "loss": 0.607,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.477164203259725e-07,
+      "loss": 0.6373,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.391726250991717e-07,
+      "loss": 0.6439,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.306673345033545e-07,
+      "loss": 0.7602,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.222005518444876e-07,
+      "loss": 0.6617,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.137722804136051e-07,
+      "loss": 0.7131,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.053825234867308e-07,
+      "loss": 0.6566,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.970312843249562e-07,
+      "loss": 0.5982,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.887185661743624e-07,
+      "loss": 0.5235,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.80444372266076e-07,
+      "loss": 0.6049,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.722087058162576e-07,
+      "loss": 0.6793,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.64011570026102e-07,
+      "loss": 0.7096,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.558529680817828e-07,
+      "loss": 0.7527,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.4773290315453e-07,
+      "loss": 0.8099,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.396513784005855e-07,
+      "loss": 0.6856,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.316083969612253e-07,
+      "loss": 0.6637,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.236039619627268e-07,
+      "loss": 0.5687,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.156380765163785e-07,
+      "loss": 0.6844,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.077107437185039e-07,
+      "loss": 0.5362,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.99821966650427e-07,
+      "loss": 0.7866,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.919717483784839e-07,
+      "loss": 0.6818,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.841600919540449e-07,
+      "loss": 0.6378,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.76387000413481e-07,
+      "loss": 0.7858,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.686524767781422e-07,
+      "loss": 0.5337,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.609565240544458e-07,
+      "loss": 0.7089,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.532991452337434e-07,
+      "loss": 0.7163,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.456803432924764e-07,
+      "loss": 0.6254,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.381001211920202e-07,
+      "loss": 0.5594,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.305584818787958e-07,
+      "loss": 0.7291,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.23055428284225e-07,
+      "loss": 0.7186,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.15590963324686e-07,
+      "loss": 0.6573,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.081650899016246e-07,
+      "loss": 0.7414,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.007778109014429e-07,
+      "loss": 0.6563,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.934291291955664e-07,
+      "loss": 0.7111,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.861190476403768e-07,
+      "loss": 0.6798,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.788475690773121e-07,
+      "loss": 0.6559,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.716146963327453e-07,
+      "loss": 0.7216,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.644204322180824e-07,
+      "loss": 0.6251,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.572647795297204e-07,
+      "loss": 0.8124,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.501477410490231e-07,
+      "loss": 0.624,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.430693195423664e-07,
+      "loss": 0.6888,
+      "step": 5013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.360295177611164e-07,
+      "loss": 0.6563,
+      "step": 5014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.290283384416063e-07,
+      "loss": 0.7973,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.220657843051924e-07,
+      "loss": 0.6785,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.15141858058188e-07,
+      "loss": 0.6716,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.082565623918956e-07,
+      "loss": 0.6844,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.014098999826301e-07,
+      "loss": 0.5971,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.946018734916293e-07,
+      "loss": 0.6965,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.878324855651873e-07,
+      "loss": 0.7268,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.811017388345331e-07,
+      "loss": 0.637,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.744096359158846e-07,
+      "loss": 0.6591,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.677561794104281e-07,
+      "loss": 0.6144,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.611413719043613e-07,
+      "loss": 0.6834,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.545652159688275e-07,
+      "loss": 0.7048,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.480277141599599e-07,
+      "loss": 0.6389,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.415288690188813e-07,
+      "loss": 0.6744,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.350686830716378e-07,
+      "loss": 0.754,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.286471588293096e-07,
+      "loss": 0.6855,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.222642987879334e-07,
+      "loss": 0.7601,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.159201054284691e-07,
+      "loss": 0.5883,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.096145812169106e-07,
+      "loss": 0.6774,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.033477286041865e-07,
+      "loss": 0.7379,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.971195500262038e-07,
+      "loss": 0.6854,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.909300479038481e-07,
+      "loss": 0.624,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.847792246429394e-07,
+      "loss": 0.6911,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.786670826342987e-07,
+      "loss": 0.6649,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7259362425369213e-07,
+      "loss": 0.6418,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6655885186185357e-07,
+      "loss": 0.6908,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6056276780447327e-07,
+      "loss": 0.7638,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.546053744122203e-07,
+      "loss": 0.7226,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4868667400070895e-07,
+      "loss": 0.5672,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.428066688705101e-07,
+      "loss": 0.7183,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.369653613071845e-07,
+      "loss": 0.7341,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3116275358121573e-07,
+      "loss": 0.7088,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.253988479480664e-07,
+      "loss": 0.5979,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1967364664814436e-07,
+      "loss": 0.6199,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1398715190681394e-07,
+      "loss": 0.6685,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.083393659343959e-07,
+      "loss": 0.6843,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0273029092617875e-07,
+      "loss": 0.7431,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.971599290623851e-07,
+      "loss": 0.6384,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.916282825081941e-07,
+      "loss": 0.6417,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.861353534137524e-07,
+      "loss": 0.627,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.806811439141189e-07,
+      "loss": 0.675,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.752656561293644e-07,
+      "loss": 0.6633,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6988889216443836e-07,
+      "loss": 0.5973,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6455085410928016e-07,
+      "loss": 0.592,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.592515440387745e-07,
+      "loss": 0.8049,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5399096401275146e-07,
+      "loss": 0.8224,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.487691160759643e-07,
+      "loss": 0.6558,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4358600225814496e-07,
+      "loss": 0.5871,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384416245739486e-07,
+      "loss": 0.6646,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.333359850229756e-07,
+      "loss": 0.7039,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.28269085589783e-07,
+      "loss": 0.7311,
+      "step": 5065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.232409282438398e-07,
+      "loss": 0.7063,
+      "step": 5066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.182515149395937e-07,
+      "loss": 0.7554,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1330084761639343e-07,
+      "loss": 0.7194,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0838892819856633e-07,
+      "loss": 0.5815,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0351575859535185e-07,
+      "loss": 0.6978,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.986813407009459e-07,
+      "loss": 0.6448,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9388567639444533e-07,
+      "loss": 0.5833,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.891287675399368e-07,
+      "loss": 0.7273,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8441061598640793e-07,
+      "loss": 0.7118,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.797312235677807e-07,
+      "loss": 0.7091,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7509059210293344e-07,
+      "loss": 0.5811,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7048872339565656e-07,
+      "loss": 0.7466,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.659256192346859e-07,
+      "loss": 0.6663,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.614012813936695e-07,
+      "loss": 0.6644,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5691571163123373e-07,
+      "loss": 0.5916,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5246891169088407e-07,
+      "loss": 0.6886,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4806088330108224e-07,
+      "loss": 0.7475,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.436916281752133e-07,
+      "loss": 0.7295,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.393611480115854e-07,
+      "loss": 0.612,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3506944449346313e-07,
+      "loss": 0.6766,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3081651928900105e-07,
+      "loss": 0.5995,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2660237405131012e-07,
+      "loss": 0.8312,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.224270104184134e-07,
+      "loss": 0.6829,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.18290430013246e-07,
+      "loss": 0.7091,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1419263444369954e-07,
+      "loss": 0.7346,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1013362530256653e-07,
+      "loss": 0.7325,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0611340416757385e-07,
+      "loss": 0.8001,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0213197260136042e-07,
+      "loss": 0.6806,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9818933215151047e-07,
+      "loss": 0.6871,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.942854843505204e-07,
+      "loss": 0.5511,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.904204307157764e-07,
+      "loss": 0.6798,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8659417274964343e-07,
+      "loss": 0.6818,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8280671193935394e-07,
+      "loss": 0.7897,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7905804975708594e-07,
+      "loss": 0.7072,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7534818765994055e-07,
+      "loss": 0.6815,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.716771270899198e-07,
+      "loss": 0.7861,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6804486947397112e-07,
+      "loss": 0.8056,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.644514162239208e-07,
+      "loss": 0.6937,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6089676873654035e-07,
+      "loss": 0.8165,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5738092839351348e-07,
+      "loss": 0.6135,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5390389656142479e-07,
+      "loss": 0.618,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5046567459181538e-07,
+      "loss": 0.8243,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4706626382107181e-07,
+      "loss": 0.7963,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.437056655705593e-07,
+      "loss": 0.672,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.403838811465219e-07,
+      "loss": 0.615,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3710091184012674e-07,
+      "loss": 0.6959,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3385675892746418e-07,
+      "loss": 0.6329,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3065142366951445e-07,
+      "loss": 0.6145,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2748490731219198e-07,
+      "loss": 0.6209,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2435721108630117e-07,
+      "loss": 0.7384,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.212683362075695e-07,
+      "loss": 0.5747,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1821828387663658e-07,
+      "loss": 0.7262,
+      "step": 5117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1520705527904297e-07,
+      "loss": 0.6717,
+      "step": 5118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1223465158526347e-07,
+      "loss": 0.6824,
+      "step": 5119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0930107395062949e-07,
+      "loss": 0.6099,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0640632351544e-07,
+      "loss": 0.6709,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.035504014048616e-07,
+      "loss": 0.6334,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0073330872898412e-07,
+      "loss": 0.6648,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.795504658280941e-08,
+      "loss": 0.5639,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.52156160462292e-08,
+      "loss": 0.6092,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.251501818405838e-08,
+      "loss": 0.6031,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.985325404601285e-08,
+      "loss": 0.684,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.723032466669834e-08,
+      "loss": 0.6752,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.464623106565484e-08,
+      "loss": 0.7845,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.210097424731222e-08,
+      "loss": 0.7546,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.959455520099024e-08,
+      "loss": 0.7183,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.71269749009318e-08,
+      "loss": 0.6973,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.469823430628076e-08,
+      "loss": 0.7484,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.230833436107087e-08,
+      "loss": 0.6248,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.995727599427016e-08,
+      "loss": 0.6463,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.764506011970317e-08,
+      "loss": 0.6407,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.537168763613988e-08,
+      "loss": 0.6098,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.313715942721787e-08,
+      "loss": 0.6933,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.094147636150904e-08,
+      "loss": 0.6389,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.878463929245292e-08,
+      "loss": 0.6194,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.666664905840113e-08,
+      "loss": 0.6388,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.458750648262845e-08,
+      "loss": 0.623,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.254721237327731e-08,
+      "loss": 0.6154,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.054576752341333e-08,
+      "loss": 0.7413,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8583172710980895e-08,
+      "loss": 0.7315,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.665942869884754e-08,
+      "loss": 0.6045,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.477453623474848e-08,
+      "loss": 0.6854,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.292849605134208e-08,
+      "loss": 0.602,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.11213088661988e-08,
+      "loss": 0.7137,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9352975381734546e-08,
+      "loss": 0.6064,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.762349628532169e-08,
+      "loss": 0.7107,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5932872249178075e-08,
+      "loss": 0.8417,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.428110393046691e-08,
+      "loss": 0.7204,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.266819197120796e-08,
+      "loss": 0.6702,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.109413699835528e-08,
+      "loss": 0.7662,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.955893962371947e-08,
+      "loss": 0.6111,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8062600444034304e-08,
+      "loss": 0.6346,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6605120040923415e-08,
+      "loss": 0.6903,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.518649898090031e-08,
+      "loss": 0.6333,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3806737815379453e-08,
+      "loss": 0.5421,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.246583708067629e-08,
+      "loss": 0.6673,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1163797297996112e-08,
+      "loss": 0.775,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.990061897342299e-08,
+      "loss": 0.6895,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8676302597964158e-08,
+      "loss": 0.6131,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7490848647494506e-08,
+      "loss": 0.6233,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6344257582812107e-08,
+      "loss": 0.6854,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.523652984958268e-08,
+      "loss": 0.6932,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.416766587838403e-08,
+      "loss": 0.6376,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3137666084672707e-08,
+      "loss": 0.744,
+      "step": 5169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2146530868806239e-08,
+      "loss": 0.7163,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1194260616043118e-08,
+      "loss": 0.6525,
+      "step": 5171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0280855696520597e-08,
+      "loss": 0.7297,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.406316465276898e-09,
+      "loss": 0.8014,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57064326225121e-09,
+      "loss": 0.5735,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.77383641225038e-09,
+      "loss": 0.7218,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.015896225004426e-09,
+      "loss": 0.7077,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.2968229951221315e-09,
+      "loss": 0.6019,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6166170021021425e-09,
+      "loss": 0.695,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.97527851033297e-09,
+      "loss": 0.6465,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372807769115195e-09,
+      "loss": 0.6137,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.809205012605954e-09,
+      "loss": 0.7221,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2844704598966602e-09,
+      "loss": 0.8232,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7986043149463848e-09,
+      "loss": 0.7771,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.351606766592962e-09,
+      "loss": 0.6497,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9434779885973973e-09,
+      "loss": 0.5535,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5742181395994592e-09,
+      "loss": 0.5903,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2438273631176777e-09,
+      "loss": 0.674,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.523057875826524e-10,
+      "loss": 0.7348,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.996535263148474e-10,
+      "loss": 0.6155,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.858706775023869e-10,
+      "loss": 0.7777,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1095732425656667e-10,
+      "loss": 0.6085,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749135345563424e-10,
+      "loss": 0.7552,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.773936128163683e-11,
+      "loss": 0.8057,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9434842213339465e-11,
+      "loss": 0.686,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7675,
+      "step": 5195
+    },
+    {
+      "epoch": 1.0,
+      "step": 5195,
+      "total_flos": 3758156120981504.0,
+      "train_loss": 0.7341916143607359,
+      "train_runtime": 149044.6655,
+      "train_samples_per_second": 4.461,
+      "train_steps_per_second": 0.035
+    }
+  ],
+  "max_steps": 5195,
+  "num_train_epochs": 1,
+  "total_flos": 3758156120981504.0,
+  "trial_name": null,
+  "trial_params": null
+}