diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31206 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 5196,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.4369,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.9945,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.657,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.834,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 2.0388,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 2.8329,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 2.7056,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 2.9414,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 3.0274,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 1.8677,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 2.1644,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 2.4931,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 2.6801,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 2.6107,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.6587,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 2.3371,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 2.2287,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 2.1203,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.7938,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 1.8124,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.7871,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.6375,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 1.6552,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.4619,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.5102,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.5069,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.316,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.5136,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 1.4618,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.4443,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 1.4401,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.4161,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.4408,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.3548,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.2641,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.1608,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 1.3529,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.3531,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.314,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 1.2979,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 1.3137,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.428,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 1.3452,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.2738,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.2212,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.3221,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 1.2539,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.2484,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 1.1897,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.2854,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.2159,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 1.2089,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 1.248,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.1994,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 1.1166,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 1.3068,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.1868,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.2338,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 1.2182,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.1294,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 1.196,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 1.2429,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.243,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 1.1853,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 1.2733,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.1956,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 1.1184,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 1.2221,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.1816,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.2016,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.1312,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.262,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 1.1636,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.2148,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.1621,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.1827,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 1.2264,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.1773,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 1.2259,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 1.1839,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 1.203,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 1.1585,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 1.1868,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.1877,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 1.1493,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.2599,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 1.207,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 1.1196,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0641025641025643e-05,
+      "loss": 1.2178,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.1728,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 1.1865,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 1.1927,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 1.1775,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 1.1682,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 1.0593,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 1.159,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 1.1272,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 1.056,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 1.1371,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 1.1522,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 1.136,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.0933,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 1.1465,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 1.1375,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 1.2304,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 1.0323,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 1.1407,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.1431,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 1.1647,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.0789,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 1.1108,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 1.157,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3717948717948718e-05,
+      "loss": 1.1391,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.2069,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 1.145,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 1.1165,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 1.1514,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 1.2081,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 1.1182,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 1.1416,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 1.115,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 1.1492,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.0992,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 1.1314,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5256410256410257e-05,
+      "loss": 1.146,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.1729,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 1.152,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 1.1025,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 1.1536,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 1.0818,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 1.1829,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.1032,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 1.1014,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 1.0997,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 1.0362,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.1707,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 1.041,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.0739,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 1.0654,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 1.0699,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 1.1904,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 1.1668,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7564102564102566e-05,
+      "loss": 1.0662,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.1339,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 0.9372,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.1583,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 1.1725,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 1.0993,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.9996,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.1197,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8589743589743593e-05,
+      "loss": 1.0868,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 1.0989,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 1.1706,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.1697,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9102564102564106e-05,
+      "loss": 1.0667,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.1355,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 0.9952,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 1.1177,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 1.1333,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 1.1004,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9871794871794873e-05,
+      "loss": 1.1283,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.1545,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998057286928e-05,
+      "loss": 1.035,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992229148463e-05,
+      "loss": 1.0805,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998251558687e-05,
+      "loss": 1.1162,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996891660592e-05,
+      "loss": 1.1674,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999951432210905e-05,
+      "loss": 1.0936,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999930062408612e-05,
+      "loss": 1.0932,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999904807207348e-05,
+      "loss": 1.0798,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999875666616918e-05,
+      "loss": 1.1162,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999842640648656e-05,
+      "loss": 1.1431,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999805729315383e-05,
+      "loss": 1.1144,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999976493263145e-05,
+      "loss": 1.1193,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99997202506127e-05,
+      "loss": 1.0585,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99996716832765e-05,
+      "loss": 1.0009,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999619230641714e-05,
+      "loss": 1.1057,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956289272873e-05,
+      "loss": 1.0906,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999502669559432e-05,
+      "loss": 1.0587,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999943856115722e-05,
+      "loss": 1.0707,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999370567547008e-05,
+      "loss": 1.0883,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999929868875521e-05,
+      "loss": 1.0816,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999922292480975e-05,
+      "loss": 1.1528,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999914327574007e-05,
+      "loss": 1.0963,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999905974157712e-05,
+      "loss": 1.0836,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998972322353353e-05,
+      "loss": 1.0904,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998881018102735e-05,
+      "loss": 1.1193,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998785828860744e-05,
+      "loss": 1.0778,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999868675466436e-05,
+      "loss": 1.1466,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998583795552086e-05,
+      "loss": 1.1088,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998476951563914e-05,
+      "loss": 1.1005,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999836622274137e-05,
+      "loss": 1.0597,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998251609127465e-05,
+      "loss": 1.1294,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998133110766744e-05,
+      "loss": 1.0519,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998010727705237e-05,
+      "loss": 1.051,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997884459990502e-05,
+      "loss": 1.0489,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997754307671597e-05,
+      "loss": 1.1611,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999762027079909e-05,
+      "loss": 1.0814,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999748234942507e-05,
+      "loss": 1.1201,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999734054360311e-05,
+      "loss": 1.0482,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997194853388316e-05,
+      "loss": 1.0993,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997045278837298e-05,
+      "loss": 1.0462,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996891820008165e-05,
+      "loss": 0.9646,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996734476960547e-05,
+      "loss": 1.0883,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996573249755573e-05,
+      "loss": 1.0324,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996408138455896e-05,
+      "loss": 1.0771,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999623914312566e-05,
+      "loss": 1.07,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996066263830533e-05,
+      "loss": 1.1024,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999588950063768e-05,
+      "loss": 1.092,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995708853615783e-05,
+      "loss": 1.0555,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995524322835035e-05,
+      "loss": 1.0969,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995335908367132e-05,
+      "loss": 1.0876,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995143610285275e-05,
+      "loss": 1.0462,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994947428664194e-05,
+      "loss": 1.1242,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99947473635801e-05,
+      "loss": 1.1269,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994543415110733e-05,
+      "loss": 0.9974,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994335583335336e-05,
+      "loss": 1.0579,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994123868334655e-05,
+      "loss": 1.0974,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 1.0887,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999368878898801e-05,
+      "loss": 1.0951,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999346542481109e-05,
+      "loss": 0.9556,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999323817774698e-05,
+      "loss": 1.0999,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993007047883988e-05,
+      "loss": 1.0776,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992772035311902e-05,
+      "loss": 1.1267,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992533140122044e-05,
+      "loss": 1.1127,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992290362407232e-05,
+      "loss": 1.0261,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992043702261795e-05,
+      "loss": 1.0748,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999179315978157e-05,
+      "loss": 1.1282,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991538735063906e-05,
+      "loss": 1.0303,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999128042820766e-05,
+      "loss": 1.1323,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999101823931319e-05,
+      "loss": 1.1109,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999075216848237e-05,
+      "loss": 1.0665,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999048221581858e-05,
+      "loss": 1.1009,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990208381426708e-05,
+      "loss": 1.0923,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989930665413148e-05,
+      "loss": 1.1422,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989649067885808e-05,
+      "loss": 1.0988,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99893635889541e-05,
+      "loss": 1.1322,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989074228728942e-05,
+      "loss": 1.0163,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998878098732277e-05,
+      "loss": 1.0675,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998848386484951e-05,
+      "loss": 1.0181,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988182861424613e-05,
+      "loss": 1.153,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998787797716503e-05,
+      "loss": 1.0713,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987569212189224e-05,
+      "loss": 1.0703,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987256566617165e-05,
+      "loss": 1.1011,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986940040570323e-05,
+      "loss": 1.0996,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986619634171682e-05,
+      "loss": 1.0507,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986295347545738e-05,
+      "loss": 1.0988,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985967180818493e-05,
+      "loss": 1.0195,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985635134117445e-05,
+      "loss": 1.0313,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985299207571614e-05,
+      "loss": 1.1259,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998495940131152e-05,
+      "loss": 1.1123,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998461571546919e-05,
+      "loss": 1.1105,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998426815017817e-05,
+      "loss": 1.1278,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998391670557349e-05,
+      "loss": 1.0242,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998356138179171e-05,
+      "loss": 1.0499,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998320217897089e-05,
+      "loss": 1.1095,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982839097250588e-05,
+      "loss": 1.0896,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998247213677188e-05,
+      "loss": 1.0471,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998210129767735e-05,
+      "loss": 1.0053,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981726580111087e-05,
+      "loss": 1.0837,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998134798421867e-05,
+      "loss": 1.0422,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980965510147214e-05,
+      "loss": 1.0695,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980579158045322e-05,
+      "loss": 1.0717,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980188928063103e-05,
+      "loss": 1.0925,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979794820352187e-05,
+      "loss": 1.0423,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979396835065696e-05,
+      "loss": 1.0724,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978994972358265e-05,
+      "loss": 1.0557,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978589232386036e-05,
+      "loss": 1.0413,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978179615306658e-05,
+      "loss": 1.1236,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997776612127928e-05,
+      "loss": 1.1141,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997734875046456e-05,
+      "loss": 1.0791,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976927503024677e-05,
+      "loss": 1.0299,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997650237912329e-05,
+      "loss": 1.0848,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976073378925586e-05,
+      "loss": 1.0299,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 1.1196,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997520375030946e-05,
+      "loss": 1.0786,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974763122228927e-05,
+      "loss": 1.0191,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997431861852785e-05,
+      "loss": 1.0851,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973870239378938e-05,
+      "loss": 1.0736,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973417984956404e-05,
+      "loss": 1.1072,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997296185543597e-05,
+      "loss": 1.0708,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972501850994857e-05,
+      "loss": 1.0944,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972037971811802e-05,
+      "loss": 1.0591,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997157021806704e-05,
+      "loss": 1.0433,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971098589942314e-05,
+      "loss": 1.0182,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970623087620867e-05,
+      "loss": 1.0677,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997014371128746e-05,
+      "loss": 1.0882,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996966046112834e-05,
+      "loss": 1.1329,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969173337331283e-05,
+      "loss": 0.9726,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968682340085546e-05,
+      "loss": 1.0568,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996818746958191e-05,
+      "loss": 1.0301,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996768872601265e-05,
+      "loss": 1.1215,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996718610957155e-05,
+      "loss": 1.1035,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966679620453903e-05,
+      "loss": 1.1023,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966169258856488e-05,
+      "loss": 1.0293,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965655024977617e-05,
+      "loss": 1.1206,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996513691901708e-05,
+      "loss": 1.0187,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964614941176194e-05,
+      "loss": 1.041,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996408909165776e-05,
+      "loss": 1.0611,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99635593706661e-05,
+      "loss": 1.0993,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996302577840703e-05,
+      "loss": 1.0655,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962488315087872e-05,
+      "loss": 0.9491,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 1.0685,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961401776106113e-05,
+      "loss": 1.0962,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996085270086568e-05,
+      "loss": 1.0535,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996029975540949e-05,
+      "loss": 1.0792,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959742939952393e-05,
+      "loss": 1.0085,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995918225471073e-05,
+      "loss": 1.0472,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958617699902354e-05,
+      "loss": 1.0905,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958049275746617e-05,
+      "loss": 1.0576,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957476982464382e-05,
+      "loss": 1.0975,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956900820278e-05,
+      "loss": 1.0719,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956320789411338e-05,
+      "loss": 0.9816,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955736890089768e-05,
+      "loss": 1.0688,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995514912254015e-05,
+      "loss": 0.9453,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995455748699087e-05,
+      "loss": 1.1257,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953961983671792e-05,
+      "loss": 1.0394,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953362612814294e-05,
+      "loss": 1.0491,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952759374651266e-05,
+      "loss": 1.1042,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952152269417088e-05,
+      "loss": 1.0449,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951541297347642e-05,
+      "loss": 1.1066,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995092645868032e-05,
+      "loss": 1.0813,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950307753654016e-05,
+      "loss": 1.0804,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949685182509117e-05,
+      "loss": 1.0644,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949058745487524e-05,
+      "loss": 1.0975,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948428442832633e-05,
+      "loss": 1.1126,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994779427478934e-05,
+      "loss": 1.0144,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994715624160405e-05,
+      "loss": 1.0399,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994651434352467e-05,
+      "loss": 1.0428,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945868580800593e-05,
+      "loss": 1.0875,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 1.0898,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944565462423502e-05,
+      "loss": 1.0505,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99439081072768e-05,
+      "loss": 1.0906,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994324688849804e-05,
+      "loss": 1.0597,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994258180634414e-05,
+      "loss": 1.088,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941912861073507e-05,
+      "loss": 1.0607,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941240052946055e-05,
+      "loss": 1.0569,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940563382223196e-05,
+      "loss": 1.0704,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9939882849167853e-05,
+      "loss": 1.0925,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993919845404444e-05,
+      "loss": 1.0162,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9938510197118868e-05,
+      "loss": 1.1305,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937818078658556e-05,
+      "loss": 1.0594,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 1.0808,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936422258210894e-05,
+      "loss": 1.0576,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935718556765878e-05,
+      "loss": 1.0578,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993501099487079e-05,
+      "loss": 1.1031,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934299572800556e-05,
+      "loss": 1.0189,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933584290831593e-05,
+      "loss": 1.0849,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932865149241813e-05,
+      "loss": 1.0927,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932142148310638e-05,
+      "loss": 1.0825,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931415288318985e-05,
+      "loss": 1.0042,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930684569549265e-05,
+      "loss": 1.113,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929949992285397e-05,
+      "loss": 0.9806,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929211556812794e-05,
+      "loss": 1.0198,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928469263418376e-05,
+      "loss": 1.1251,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927723112390547e-05,
+      "loss": 1.0497,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992697310401922e-05,
+      "loss": 0.9944,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992621923859581e-05,
+      "loss": 1.0336,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925461516413224e-05,
+      "loss": 1.0151,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924699937765863e-05,
+      "loss": 1.0945,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923934502949645e-05,
+      "loss": 1.1118,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992316521226196e-05,
+      "loss": 1.0473,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922392066001724e-05,
+      "loss": 1.0687,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921615064469328e-05,
+      "loss": 1.0628,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992083420796667e-05,
+      "loss": 1.0524,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920049496797153e-05,
+      "loss": 1.0896,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919260931265666e-05,
+      "loss": 1.0816,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99184685116786e-05,
+      "loss": 1.1124,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991767223834384e-05,
+      "loss": 1.0809,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916872111570782e-05,
+      "loss": 1.042,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916068131670302e-05,
+      "loss": 1.067,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915260298954786e-05,
+      "loss": 1.0687,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914448613738107e-05,
+      "loss": 1.0989,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913633076335638e-05,
+      "loss": 1.0794,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912813687064255e-05,
+      "loss": 1.0705,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991199044624232e-05,
+      "loss": 1.1429,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911163354189702e-05,
+      "loss": 1.0228,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991033241122776e-05,
+      "loss": 1.0634,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990949761767935e-05,
+      "loss": 1.1184,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908658973868823e-05,
+      "loss": 1.0725,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907816480122032e-05,
+      "loss": 0.9799,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906970136766323e-05,
+      "loss": 1.0444,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906119944130527e-05,
+      "loss": 1.0964,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990526590254499e-05,
+      "loss": 1.0176,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904408012341542e-05,
+      "loss": 0.9376,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903546273853503e-05,
+      "loss": 1.1214,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 1.0021,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901811253364458e-05,
+      "loss": 1.0524,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900937972037577e-05,
+      "loss": 1.0095,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900060843774364e-05,
+      "loss": 1.0807,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989917986891563e-05,
+      "loss": 1.0597,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9898295047803665e-05,
+      "loss": 1.0668,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9897406380782262e-05,
+      "loss": 1.1265,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9896513868196706e-05,
+      "loss": 1.079,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9895617510393773e-05,
+      "loss": 1.0614,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894717307721734e-05,
+      "loss": 1.0378,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893813260530368e-05,
+      "loss": 1.0091,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989290536917093e-05,
+      "loss": 1.0347,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891993633996164e-05,
+      "loss": 1.0004,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891078055360332e-05,
+      "loss": 1.0558,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989015863361917e-05,
+      "loss": 1.0103,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889235369129913e-05,
+      "loss": 0.986,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 1.0976,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887377313343514e-05,
+      "loss": 1.0717,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988644252276831e-05,
+      "loss": 1.0881,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885503890888874e-05,
+      "loss": 1.0699,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988456141806991e-05,
+      "loss": 1.0135,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988361510467761e-05,
+      "loss": 1.0954,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882664951079652e-05,
+      "loss": 1.075,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988171095764521e-05,
+      "loss": 1.1002,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880753124744964e-05,
+      "loss": 1.092,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879791452751055e-05,
+      "loss": 1.0048,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878825942037147e-05,
+      "loss": 1.0805,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987785659297838e-05,
+      "loss": 1.0337,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 1.0191,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875906381334277e-05,
+      "loss": 1.1057,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874925519506686e-05,
+      "loss": 1.0991,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873940820849714e-05,
+      "loss": 1.0917,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872952285745958e-05,
+      "loss": 1.0558,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987195991457951e-05,
+      "loss": 1.0466,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987096370773594e-05,
+      "loss": 1.078,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869963665602322e-05,
+      "loss": 0.9763,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868959788567213e-05,
+      "loss": 1.0282,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867952077020667e-05,
+      "loss": 1.0018,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866940531354214e-05,
+      "loss": 1.0548,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865925151960893e-05,
+      "loss": 1.105,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864905939235215e-05,
+      "loss": 1.0776,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863882893573188e-05,
+      "loss": 1.0123,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862856015372315e-05,
+      "loss": 1.0789,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861825305031576e-05,
+      "loss": 1.0313,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860790762951447e-05,
+      "loss": 1.0234,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985975238953389e-05,
+      "loss": 1.0507,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985871018518236e-05,
+      "loss": 1.0296,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857664150301798e-05,
+      "loss": 1.0121,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985661428529863e-05,
+      "loss": 1.0265,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985556059058078e-05,
+      "loss": 1.0527,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854503066557646e-05,
+      "loss": 1.0362,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853441713640123e-05,
+      "loss": 1.031,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852376532240594e-05,
+      "loss": 1.0307,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851307522772926e-05,
+      "loss": 0.9767,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850234685652478e-05,
+      "loss": 1.0356,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849158021296083e-05,
+      "loss": 1.0459,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 1.0946,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9846993212550286e-05,
+      "loss": 1.0558,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9845905069002004e-05,
+      "loss": 1.0054,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984481309990002e-05,
+      "loss": 1.0951,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984371730566861e-05,
+      "loss": 0.9907,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842617686733546e-05,
+      "loss": 1.0039,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841514243522065e-05,
+      "loss": 1.0009,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984040697646291e-05,
+      "loss": 1.0692,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98392958859863e-05,
+      "loss": 1.017,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838180972523936e-05,
+      "loss": 1.1193,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837062236509013e-05,
+      "loss": 1.0122,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835939678376213e-05,
+      "loss": 0.9964,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834813298561686e-05,
+      "loss": 0.9924,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833683097503087e-05,
+      "loss": 1.1566,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983254907563955e-05,
+      "loss": 1.11,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983141123341168e-05,
+      "loss": 1.1083,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830269571261585e-05,
+      "loss": 1.0498,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982912408963285e-05,
+      "loss": 1.1104,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827974788970536e-05,
+      "loss": 0.9953,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826821669721205e-05,
+      "loss": 1.0593,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825664732332886e-05,
+      "loss": 1.0752,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98245039772551e-05,
+      "loss": 1.0213,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982333940493885e-05,
+      "loss": 0.9984,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822171015836625e-05,
+      "loss": 1.0028,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982099881040239e-05,
+      "loss": 1.1234,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819822789091597e-05,
+      "loss": 1.0533,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818642952361188e-05,
+      "loss": 1.0037,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817459300669567e-05,
+      "loss": 1.0764,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816271834476642e-05,
+      "loss": 1.0157,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981508055424379e-05,
+      "loss": 0.9857,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981388546043388e-05,
+      "loss": 1.0088,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812686553511246e-05,
+      "loss": 1.06,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811483833941726e-05,
+      "loss": 1.0676,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810277302192624e-05,
+      "loss": 1.0128,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809066958732727e-05,
+      "loss": 1.0532,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 1.048,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806634838563113e-05,
+      "loss": 1.0132,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805413062798376e-05,
+      "loss": 1.0074,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980418747721281e-05,
+      "loss": 1.0886,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980295808228261e-05,
+      "loss": 1.0435,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801724878485438e-05,
+      "loss": 1.133,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980048786630046e-05,
+      "loss": 1.1239,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799247046208297e-05,
+      "loss": 1.0876,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798002418691067e-05,
+      "loss": 1.0538,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796753984232357e-05,
+      "loss": 1.1043,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979550174331724e-05,
+      "loss": 1.0255,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794245696432262e-05,
+      "loss": 0.9719,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979298584406545e-05,
+      "loss": 0.9443,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791722186706318e-05,
+      "loss": 0.9897,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979045472484584e-05,
+      "loss": 0.9826,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789183458976485e-05,
+      "loss": 1.0959,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787908389592197e-05,
+      "loss": 1.0002,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9786629517188384e-05,
+      "loss": 1.0571,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9785346842261957e-05,
+      "loss": 1.0367,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9784060365311276e-05,
+      "loss": 1.0458,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97827700868362e-05,
+      "loss": 1.0449,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 1.0191,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.978017812731965e-05,
+      "loss": 1.0966,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778876447285264e-05,
+      "loss": 1.0354,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777570967740648e-05,
+      "loss": 1.0243,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977626168919305e-05,
+      "loss": 1.0644,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977494861215117e-05,
+      "loss": 0.9727,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773631737125192e-05,
+      "loss": 1.0115,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977231106462679e-05,
+      "loss": 1.0261,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770986595169096e-05,
+      "loss": 1.0584,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769658329266718e-05,
+      "loss": 0.9809,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976832626743575e-05,
+      "loss": 1.0684,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766990410193755e-05,
+      "loss": 0.9584,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765650758059766e-05,
+      "loss": 1.0134,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97643073115543e-05,
+      "loss": 1.0743,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762960071199334e-05,
+      "loss": 1.0776,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976160903751834e-05,
+      "loss": 0.9883,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760254211036245e-05,
+      "loss": 1.0474,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758895592279458e-05,
+      "loss": 1.0754,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757533181775865e-05,
+      "loss": 1.033,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756166980054812e-05,
+      "loss": 1.0003,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754796987647134e-05,
+      "loss": 1.0289,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753423205085127e-05,
+      "loss": 0.993,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752045632902567e-05,
+      "loss": 1.0368,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97506642716347e-05,
+      "loss": 1.0676,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 1.0834,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747890183991376e-05,
+      "loss": 1.0955,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746497458693776e-05,
+      "loss": 0.9249,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745100946466563e-05,
+      "loss": 1.0486,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743700647852356e-05,
+      "loss": 1.0547,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742296563395218e-05,
+      "loss": 1.0299,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740888693640698e-05,
+      "loss": 1.0293,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973947703913582e-05,
+      "loss": 1.0537,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738061600429062e-05,
+      "loss": 1.0369,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736642378070393e-05,
+      "loss": 1.1023,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735219372611232e-05,
+      "loss": 1.0441,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733792584604487e-05,
+      "loss": 1.0796,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732362014604515e-05,
+      "loss": 1.0707,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973092766316716e-05,
+      "loss": 1.0714,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972948953084973e-05,
+      "loss": 0.929,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728047618210995e-05,
+      "loss": 1.0373,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726601925811204e-05,
+      "loss": 0.9999,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725152454212068e-05,
+      "loss": 1.0441,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723699203976768e-05,
+      "loss": 1.007,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722242175669955e-05,
+      "loss": 1.0989,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720781369857747e-05,
+      "loss": 1.0308,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719316787107728e-05,
+      "loss": 0.9771,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717848427988954e-05,
+      "loss": 1.0152,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971637629307194e-05,
+      "loss": 1.082,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9714900382928674e-05,
+      "loss": 1.011,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9713420698132614e-05,
+      "loss": 1.0414,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.971193723925868e-05,
+      "loss": 1.0452,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9710450006883258e-05,
+      "loss": 1.0543,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9708959001584197e-05,
+      "loss": 1.0697,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970746422394082e-05,
+      "loss": 0.9867,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970596567453391e-05,
+      "loss": 1.0071,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704463353945724e-05,
+      "loss": 1.0287,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702957262759964e-05,
+      "loss": 0.9421,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701447401561826e-05,
+      "loss": 1.0777,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699933770937948e-05,
+      "loss": 1.0421,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698416371476434e-05,
+      "loss": 1.0787,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969689520376687e-05,
+      "loss": 1.0278,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695370268400288e-05,
+      "loss": 0.9845,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969384156596919e-05,
+      "loss": 1.1038,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692309097067546e-05,
+      "loss": 1.0376,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969077286229078e-05,
+      "loss": 1.0471,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968923286223579e-05,
+      "loss": 1.006,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687689097500927e-05,
+      "loss": 1.1064,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686141568686015e-05,
+      "loss": 1.0835,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968459027639233e-05,
+      "loss": 1.0341,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683035221222617e-05,
+      "loss": 1.01,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968147640378108e-05,
+      "loss": 1.0937,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679913824673388e-05,
+      "loss": 0.9999,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678347484506667e-05,
+      "loss": 1.0888,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676777383889506e-05,
+      "loss": 1.007,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675203523431964e-05,
+      "loss": 1.0249,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673625903745546e-05,
+      "loss": 1.0594,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672044525443222e-05,
+      "loss": 1.0338,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670459389139433e-05,
+      "loss": 0.9852,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668870495450064e-05,
+      "loss": 1.0037,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667277844992476e-05,
+      "loss": 1.0251,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665681438385475e-05,
+      "loss": 0.9855,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664081276249335e-05,
+      "loss": 1.0873,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662477359205786e-05,
+      "loss": 1.0305,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966086968787802e-05,
+      "loss": 0.9804,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 1.042,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965764308486989e-05,
+      "loss": 1.0569,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656024154443195e-05,
+      "loss": 1.064,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965440147223963e-05,
+      "loss": 1.0749,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652775038889676e-05,
+      "loss": 1.0592,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651144855025265e-05,
+      "loss": 1.033,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649510921279803e-05,
+      "loss": 0.93,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964787323828813e-05,
+      "loss": 0.9896,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646231806686566e-05,
+      "loss": 1.0594,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644586627112877e-05,
+      "loss": 1.0519,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964293770020628e-05,
+      "loss": 0.9735,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641285026607453e-05,
+      "loss": 1.0769,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639628606958535e-05,
+      "loss": 1.0766,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637968441903112e-05,
+      "loss": 1.0511,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963630453208623e-05,
+      "loss": 1.0265,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634636878154393e-05,
+      "loss": 0.9495,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.963296548075555e-05,
+      "loss": 1.0806,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.963129034053911e-05,
+      "loss": 1.0316,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9629611458155936e-05,
+      "loss": 1.0287,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962792883425835e-05,
+      "loss": 0.9968,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962624246950012e-05,
+      "loss": 0.9989,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9624552364536472e-05,
+      "loss": 0.9723,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622858520024087e-05,
+      "loss": 1.1298,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621160936621086e-05,
+      "loss": 1.0042,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961945961498706e-05,
+      "loss": 0.9917,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617754555783045e-05,
+      "loss": 1.0616,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616045759671524e-05,
+      "loss": 1.0916,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614333227316443e-05,
+      "loss": 1.0714,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 0.9511,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961089695653861e-05,
+      "loss": 1.0995,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609173219450998e-05,
+      "loss": 1.0605,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607445748790096e-05,
+      "loss": 0.9984,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605714545227105e-05,
+      "loss": 1.0542,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603979609434666e-05,
+      "loss": 1.0373,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960224094208688e-05,
+      "loss": 1.0006,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960049854385929e-05,
+      "loss": 0.9861,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598752415428893e-05,
+      "loss": 0.9544,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597002557474136e-05,
+      "loss": 1.0689,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959524897067491e-05,
+      "loss": 1.017,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959349165571256e-05,
+      "loss": 1.0287,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591730613269878e-05,
+      "loss": 1.0432,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589965844031108e-05,
+      "loss": 1.0562,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958819734868193e-05,
+      "loss": 1.0282,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586425127909488e-05,
+      "loss": 1.0066,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584649182402358e-05,
+      "loss": 0.9821,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582869512850576e-05,
+      "loss": 1.023,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581086119945615e-05,
+      "loss": 1.0217,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579299004380402e-05,
+      "loss": 1.0496,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577508166849308e-05,
+      "loss": 1.0993,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575713608048146e-05,
+      "loss": 0.9449,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957391532867418e-05,
+      "loss": 0.9995,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572113329426123e-05,
+      "loss": 0.8996,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570307611004124e-05,
+      "loss": 1.0815,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568498174109783e-05,
+      "loss": 1.0741,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956668501944614e-05,
+      "loss": 1.0756,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956486814771769e-05,
+      "loss": 1.034,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563047559630356e-05,
+      "loss": 1.0481,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956122325589152e-05,
+      "loss": 1.0645,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559395237210002e-05,
+      "loss": 1.0546,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557563504296064e-05,
+      "loss": 1.0215,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 1.022,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955388889861919e-05,
+      "loss": 1.0109,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552046027284e-05,
+      "loss": 0.9922,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550199444571868e-05,
+      "loss": 0.9953,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548349151200275e-05,
+      "loss": 0.9813,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546495147888134e-05,
+      "loss": 0.9957,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954463743535581e-05,
+      "loss": 1.0141,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95427760143251e-05,
+      "loss": 0.9736,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9540910885519243e-05,
+      "loss": 0.9556,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9539042049662926e-05,
+      "loss": 0.9884,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953716950748227e-05,
+      "loss": 1.0102,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953529325970484e-05,
+      "loss": 1.0173,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953341330705963e-05,
+      "loss": 1.073,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9531529650277087e-05,
+      "loss": 0.994,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952964229008909e-05,
+      "loss": 1.0445,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527751227228964e-05,
+      "loss": 1.0324,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525856462431463e-05,
+      "loss": 1.0012,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523957996432785e-05,
+      "loss": 0.9817,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522055829970564e-05,
+      "loss": 0.9921,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520149963783873e-05,
+      "loss": 1.0489,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518240398613226e-05,
+      "loss": 0.9979,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516327135200565e-05,
+      "loss": 0.9626,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951441017428928e-05,
+      "loss": 1.0561,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951248951662418e-05,
+      "loss": 1.0376,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 1.0541,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508637114019037e-05,
+      "loss": 1.0162,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950670537057581e-05,
+      "loss": 1.0567,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504769933372424e-05,
+      "loss": 1.0175,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502830803160873e-05,
+      "loss": 0.9943,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500887980694597e-05,
+      "loss": 1.0861,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498941466728462e-05,
+      "loss": 1.0491,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949699126201877e-05,
+      "loss": 1.0089,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495037367323264e-05,
+      "loss": 0.9673,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493079783401116e-05,
+      "loss": 1.0619,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949111851101292e-05,
+      "loss": 1.0356,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489153550920726e-05,
+      "loss": 1.0092,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487184903888e-05,
+      "loss": 1.0347,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485212570679643e-05,
+      "loss": 1.0299,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483236552061996e-05,
+      "loss": 1.0394,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948125684880282e-05,
+      "loss": 0.9794,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947927346167132e-05,
+      "loss": 0.9952,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477286391438122e-05,
+      "loss": 1.0104,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947529563887529e-05,
+      "loss": 1.0584,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473301204756315e-05,
+      "loss": 0.9556,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947130308985612e-05,
+      "loss": 1.0504,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946930129495106e-05,
+      "loss": 1.033,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467295820818913e-05,
+      "loss": 0.9566,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465286668238895e-05,
+      "loss": 0.9971,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463273837991643e-05,
+      "loss": 1.012,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461257330859233e-05,
+      "loss": 0.9727,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945923714762516e-05,
+      "loss": 1.0109,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457213289074353e-05,
+      "loss": 1.0434,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945518575599317e-05,
+      "loss": 1.0137,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945315454916939e-05,
+      "loss": 1.044,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945111966939222e-05,
+      "loss": 1.0261,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449081117452304e-05,
+      "loss": 1.04,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447038894141707e-05,
+      "loss": 0.9804,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444993000253912e-05,
+      "loss": 1.0485,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442943436583845e-05,
+      "loss": 1.0226,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440890203927846e-05,
+      "loss": 1.0389,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 1.0639,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.943677273485054e-05,
+      "loss": 1.0339,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.943470850002905e-05,
+      "loss": 1.0659,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.943264059942125e-05,
+      "loss": 1.0397,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9430569033830606e-05,
+      "loss": 1.0475,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9428493804062013e-05,
+      "loss": 1.0469,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426414910921785e-05,
+      "loss": 0.9709,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942433235521766e-05,
+      "loss": 1.0561,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94222461377588e-05,
+      "loss": 0.9591,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942015625935579e-05,
+      "loss": 1.0464,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418062720820636e-05,
+      "loss": 0.8708,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415965522966767e-05,
+      "loss": 1.055,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413864666609036e-05,
+      "loss": 0.9842,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941176015256371e-05,
+      "loss": 1.0138,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409651981648486e-05,
+      "loss": 0.9952,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407540154682473e-05,
+      "loss": 1.0205,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405424672486218e-05,
+      "loss": 1.0908,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403305535881664e-05,
+      "loss": 1.0382,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940118274569219e-05,
+      "loss": 0.9737,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399056302742598e-05,
+      "loss": 0.9651,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 1.0877,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394792461869296e-05,
+      "loss": 1.0338,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939265506560228e-05,
+      "loss": 0.9668,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390514019888508e-05,
+      "loss": 1.0595,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388369325559862e-05,
+      "loss": 1.0165,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386220983449652e-05,
+      "loss": 1.0405,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93840689943926e-05,
+      "loss": 0.9801,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381913359224844e-05,
+      "loss": 1.0832,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937975407878394e-05,
+      "loss": 0.9913,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937759115390886e-05,
+      "loss": 1.0205,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375424585439994e-05,
+      "loss": 1.0928,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373254374219146e-05,
+      "loss": 1.0308,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371080521089536e-05,
+      "loss": 1.0724,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368903026895795e-05,
+      "loss": 1.0787,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366721892483976e-05,
+      "loss": 1.0575,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364537118701542e-05,
+      "loss": 0.9749,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362348706397374e-05,
+      "loss": 1.0764,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360156656421756e-05,
+      "loss": 0.9943,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357960969626398e-05,
+      "loss": 1.0275,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355761646864417e-05,
+      "loss": 1.0303,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935355868899034e-05,
+      "loss": 1.0061,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935135209686012e-05,
+      "loss": 1.0038,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93491418713311e-05,
+      "loss": 0.9949,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346928013262052e-05,
+      "loss": 1.0024,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344710523513157e-05,
+      "loss": 1.0166,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342489402945997e-05,
+      "loss": 1.0145,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340264652423584e-05,
+      "loss": 0.9946,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338036272810314e-05,
+      "loss": 0.9403,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335804264972018e-05,
+      "loss": 1.0472,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933356862977592e-05,
+      "loss": 1.0386,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331329368090664e-05,
+      "loss": 1.0506,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93290864807863e-05,
+      "loss": 0.9953,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9326839968734278e-05,
+      "loss": 1.0297,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9324589832807467e-05,
+      "loss": 1.0398,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9322336073880143e-05,
+      "loss": 1.0118,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.932007869282799e-05,
+      "loss": 0.9715,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9317817690528086e-05,
+      "loss": 1.0291,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9315553067858933e-05,
+      "loss": 0.9854,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931328482570044e-05,
+      "loss": 1.0268,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93110129649339e-05,
+      "loss": 1.0667,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308737486442045e-05,
+      "loss": 1.0618,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306458391108984e-05,
+      "loss": 0.9984,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304175679820247e-05,
+      "loss": 1.0396,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301889353462765e-05,
+      "loss": 1.0885,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929959941292487e-05,
+      "loss": 0.9866,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297305859096305e-05,
+      "loss": 1.0218,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929500869286821e-05,
+      "loss": 1.0205,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292707915133134e-05,
+      "loss": 0.9753,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290403526785025e-05,
+      "loss": 0.9824,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288095528719245e-05,
+      "loss": 1.0259,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285783921832537e-05,
+      "loss": 1.0366,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283468707023068e-05,
+      "loss": 0.9133,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928114988519039e-05,
+      "loss": 0.9855,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278827457235473e-05,
+      "loss": 1.0614,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927650142406067e-05,
+      "loss": 1.0606,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927417178656975e-05,
+      "loss": 1.0344,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271838545667876e-05,
+      "loss": 1.0361,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926950170226161e-05,
+      "loss": 1.0595,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267161257258916e-05,
+      "loss": 0.98,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264817211569154e-05,
+      "loss": 1.0445,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926246956610309e-05,
+      "loss": 1.0064,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926011832177288e-05,
+      "loss": 0.9791,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257763479492086e-05,
+      "loss": 1.0362,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255405040175666e-05,
+      "loss": 1.0327,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253043004739967e-05,
+      "loss": 1.0159,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250677374102752e-05,
+      "loss": 1.0372,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248308149183155e-05,
+      "loss": 0.9646,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245935330901733e-05,
+      "loss": 1.0553,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243558920180417e-05,
+      "loss": 1.0868,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241178917942552e-05,
+      "loss": 0.9709,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 1.0781,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923640814261749e-05,
+      "loss": 1.035,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234017371383946e-05,
+      "loss": 1.076,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231623012341145e-05,
+      "loss": 1.0199,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92292250664194e-05,
+      "loss": 1.0289,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226823534550418e-05,
+      "loss": 0.9996,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224418417667295e-05,
+      "loss": 1.0501,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922200971670452e-05,
+      "loss": 0.9664,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219597432597975e-05,
+      "loss": 1.098,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217181566284942e-05,
+      "loss": 1.072,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921476211870408e-05,
+      "loss": 1.059,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921233909079545e-05,
+      "loss": 1.0337,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92099124835005e-05,
+      "loss": 0.9669,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207482297762076e-05,
+      "loss": 1.0386,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9205048534524405e-05,
+      "loss": 1.016,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9202611194733107e-05,
+      "loss": 1.0084,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9200170279335194e-05,
+      "loss": 1.0875,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9197725789279065e-05,
+      "loss": 1.0558,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919527772551451e-05,
+      "loss": 0.9138,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91928260889927e-05,
+      "loss": 1.0256,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190370880666206e-05,
+      "loss": 1.0453,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187912101488986e-05,
+      "loss": 0.9968,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185449752416367e-05,
+      "loss": 0.8893,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182983834405088e-05,
+      "loss": 0.9755,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918051434841326e-05,
+      "loss": 0.9896,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178041295400383e-05,
+      "loss": 1.0253,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917556467632734e-05,
+      "loss": 1.0569,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173084492156406e-05,
+      "loss": 1.0529,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917060074385124e-05,
+      "loss": 1.0128,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168113432376885e-05,
+      "loss": 1.0209,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165622558699763e-05,
+      "loss": 1.0846,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163128123787688e-05,
+      "loss": 1.029,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916063012860985e-05,
+      "loss": 1.0003,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158128574136833e-05,
+      "loss": 0.9772,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155623461340594e-05,
+      "loss": 0.9632,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153114791194475e-05,
+      "loss": 1.0186,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91506025646732e-05,
+      "loss": 0.9905,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914808678275288e-05,
+      "loss": 1.074,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145567446411003e-05,
+      "loss": 1.028,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143044556626437e-05,
+      "loss": 1.0296,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140518114379433e-05,
+      "loss": 0.9756,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913798812065162e-05,
+      "loss": 1.0238,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 1.0523,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913291748268699e-05,
+      "loss": 1.0521,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130376840420337e-05,
+      "loss": 1.0223,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912783265061319e-05,
+      "loss": 1.0514,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912528491425408e-05,
+      "loss": 0.9819,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122733632332907e-05,
+      "loss": 1.0634,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120178805840956e-05,
+      "loss": 1.0176,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911762043577089e-05,
+      "loss": 0.9855,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115058523116734e-05,
+      "loss": 1.0499,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112493068873907e-05,
+      "loss": 1.0456,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109924074039198e-05,
+      "loss": 1.0523,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910735153961077e-05,
+      "loss": 0.9821,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104775466588162e-05,
+      "loss": 1.0243,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102195855972287e-05,
+      "loss": 0.9611,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099612708765432e-05,
+      "loss": 0.9936,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097026025971268e-05,
+      "loss": 0.9876,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094435808594823e-05,
+      "loss": 1.0959,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909184205764251e-05,
+      "loss": 1.0254,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908924477412211e-05,
+      "loss": 1.03,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908664395904278e-05,
+      "loss": 0.9732,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084039613415052e-05,
+      "loss": 0.9926,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081431738250815e-05,
+      "loss": 1.0253,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078820334563346e-05,
+      "loss": 0.9928,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076205403367287e-05,
+      "loss": 1.0607,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.907358694567865e-05,
+      "loss": 0.9193,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9070964962514816e-05,
+      "loss": 1.0507,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9068339454894536e-05,
+      "loss": 1.019,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9065710423837936e-05,
+      "loss": 1.0055,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9063077870366504e-05,
+      "loss": 1.0474,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9060441795503095e-05,
+      "loss": 1.0119,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057802200271943e-05,
+      "loss": 1.0822,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055159085698638e-05,
+      "loss": 1.0226,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905251245281015e-05,
+      "loss": 1.0996,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90498623026348e-05,
+      "loss": 1.0389,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904720863620229e-05,
+      "loss": 0.9806,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044551454543683e-05,
+      "loss": 1.0537,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041890758691403e-05,
+      "loss": 1.0091,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039226549679243e-05,
+      "loss": 1.0464,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903655882854237e-05,
+      "loss": 1.0064,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90338875963173e-05,
+      "loss": 1.0035,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903121285404192e-05,
+      "loss": 1.0609,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028534602755485e-05,
+      "loss": 1.0168,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902585284349861e-05,
+      "loss": 0.9842,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023167577313267e-05,
+      "loss": 1.0307,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020478805242803e-05,
+      "loss": 1.0175,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901778652833192e-05,
+      "loss": 0.9949,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901509074762668e-05,
+      "loss": 1.0392,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901239146417451e-05,
+      "loss": 1.0885,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.9766,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006982393225878e-05,
+      "loss": 1.0606,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004272607831077e-05,
+      "loss": 1.0111,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001559323892652e-05,
+      "loss": 1.0546,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998842542464833e-05,
+      "loss": 1.0365,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996122264603202e-05,
+      "loss": 0.9703,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993398491364702e-05,
+      "loss": 0.9787,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899067122380764e-05,
+      "loss": 1.1028,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987940462991673e-05,
+      "loss": 0.9787,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985206209977813e-05,
+      "loss": 1.0097,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898246846582844e-05,
+      "loss": 1.0539,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897972723160729e-05,
+      "loss": 1.0066,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976982508379436e-05,
+      "loss": 0.955,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974234297211327e-05,
+      "loss": 1.0266,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897148259917076e-05,
+      "loss": 1.0683,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968727415326885e-05,
+      "loss": 1.0373,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896596874675021e-05,
+      "loss": 1.0359,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963206594512593e-05,
+      "loss": 0.9679,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960440959687254e-05,
+      "loss": 0.982,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957671843348754e-05,
+      "loss": 1.0024,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895489924657301e-05,
+      "loss": 1.0093,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952123170437304e-05,
+      "loss": 0.97,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894934361602025e-05,
+      "loss": 1.007,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946560584401833e-05,
+      "loss": 1.0573,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943774076663372e-05,
+      "loss": 1.003,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894098409388754e-05,
+      "loss": 0.9726,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938190637158377e-05,
+      "loss": 1.032,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893539370756125e-05,
+      "loss": 1.0765,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.893259330618289e-05,
+      "loss": 1.0248,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.892978943411137e-05,
+      "loss": 0.963,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8926982092436117e-05,
+      "loss": 1.0232,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89241712822479e-05,
+      "loss": 1.0732,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8921357004638837e-05,
+      "loss": 0.9373,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8918539260702394e-05,
+      "loss": 1.0629,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891571805153339e-05,
+      "loss": 1.0515,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912893378227984e-05,
+      "loss": 0.994,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.9531,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890723364359933e-05,
+      "loss": 1.0078,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890439858447513e-05,
+      "loss": 1.0408,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890156006561262e-05,
+      "loss": 1.07,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898718088114688e-05,
+      "loss": 0.9858,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895872653085563e-05,
+      "loss": 1.0185,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893023761630816e-05,
+      "loss": 1.0137,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890171414857366e-05,
+      "loss": 0.998,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888731561387347e-05,
+      "loss": 0.9691,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884456359788725e-05,
+      "loss": 0.9709,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881593653714075e-05,
+      "loss": 1.0291,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878727496761808e-05,
+      "loss": 1.0031,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875857890045544e-05,
+      "loss": 1.0214,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872984834680242e-05,
+      "loss": 1.0028,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887010833178222e-05,
+      "loss": 1.0191,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886722838246911e-05,
+      "loss": 1.0403,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864344987859898e-05,
+      "loss": 1.0143,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886145814907491e-05,
+      "loss": 0.9589,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88585678672358e-05,
+      "loss": 1.0481,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855674143465567e-05,
+      "loss": 1.043,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885277697888855e-05,
+      "loss": 0.9766,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884987637463042e-05,
+      "loss": 1.0549,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846972331818182e-05,
+      "loss": 1.0102,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844064851580182e-05,
+      "loss": 1.0132,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841153935046098e-05,
+      "loss": 1.0224,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838239583346948e-05,
+      "loss": 1.0654,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835321797615083e-05,
+      "loss": 1.0435,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832400578984183e-05,
+      "loss": 1.0032,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 0.9886,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826547847566692e-05,
+      "loss": 1.0292,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823616337054136e-05,
+      "loss": 1.0283,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820681398190618e-05,
+      "loss": 0.9484,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817743032116483e-05,
+      "loss": 1.026,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814801239973413e-05,
+      "loss": 1.0464,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811856022904423e-05,
+      "loss": 1.0325,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880890738205386e-05,
+      "loss": 1.0439,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880595531856738e-05,
+      "loss": 1.0801,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802999833592e-05,
+      "loss": 1.0451,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880004092827605e-05,
+      "loss": 0.9799,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797078603769184e-05,
+      "loss": 1.0263,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87941128612224e-05,
+      "loss": 1.0026,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791143701788004e-05,
+      "loss": 1.0418,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788171126619653e-05,
+      "loss": 1.0259,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785195136872315e-05,
+      "loss": 1.0633,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8782215733702286e-05,
+      "loss": 0.9988,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8779232918267194e-05,
+      "loss": 1.0154,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8776246691725988e-05,
+      "loss": 1.0279,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8773257055238946e-05,
+      "loss": 0.9657,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8770264009967667e-05,
+      "loss": 0.9589,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876726755707508e-05,
+      "loss": 0.9235,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764267697725432e-05,
+      "loss": 1.0208,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761264433084294e-05,
+      "loss": 0.9803,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758257764318566e-05,
+      "loss": 0.9823,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755247692596467e-05,
+      "loss": 1.0704,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752234219087538e-05,
+      "loss": 0.9879,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874921734496264e-05,
+      "loss": 1.0378,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874619707139396e-05,
+      "loss": 1.0393,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743173399555e-05,
+      "loss": 1.0564,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740146330620584e-05,
+      "loss": 0.9562,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737115865766865e-05,
+      "loss": 1.0163,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87340820061713e-05,
+      "loss": 0.9837,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731044753012674e-05,
+      "loss": 0.9831,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872800410747109e-05,
+      "loss": 1.0145,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724960070727974e-05,
+      "loss": 1.0405,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721912643966055e-05,
+      "loss": 1.0316,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871886182836939e-05,
+      "loss": 1.0174,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871580762512336e-05,
+      "loss": 0.9959,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871275003541464e-05,
+      "loss": 1.032,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709689060431242e-05,
+      "loss": 1.0421,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706624701362485e-05,
+      "loss": 1.0416,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703556959398998e-05,
+      "loss": 0.8839,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700485835732735e-05,
+      "loss": 1.0857,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697411331556958e-05,
+      "loss": 0.9524,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694333448066238e-05,
+      "loss": 1.0088,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691252186456465e-05,
+      "loss": 0.9782,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688167547924846e-05,
+      "loss": 0.9769,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868507953366989e-05,
+      "loss": 0.9126,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681988144891425e-05,
+      "loss": 1.0453,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678893382790583e-05,
+      "loss": 1.0279,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675795248569816e-05,
+      "loss": 1.0029,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672693743432875e-05,
+      "loss": 1.0257,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669588868584832e-05,
+      "loss": 0.9749,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866648062523206e-05,
+      "loss": 1.0515,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866336901458225e-05,
+      "loss": 1.0513,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 1.1206,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657135696228777e-05,
+      "loss": 1.0424,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865401399094703e-05,
+      "loss": 0.9657,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650888923212057e-05,
+      "loss": 0.9782,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647760494238082e-05,
+      "loss": 1.0529,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644628705240636e-05,
+      "loss": 1.0387,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864149355743655e-05,
+      "loss": 0.9816,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863835505204396e-05,
+      "loss": 1.0889,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635213190282312e-05,
+      "loss": 1.0082,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863206797337235e-05,
+      "loss": 1.0634,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862891940253613e-05,
+      "loss": 1.0548,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625767478997003e-05,
+      "loss": 1.0574,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8622612203979628e-05,
+      "loss": 1.0391,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8619453578709957e-05,
+      "loss": 0.9787,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.861629160441526e-05,
+      "loss": 0.9659,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8613126282324092e-05,
+      "loss": 1.0652,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8609957613666316e-05,
+      "loss": 0.9702,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8606785599673098e-05,
+      "loss": 0.9993,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.86036102415769e-05,
+      "loss": 1.0237,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860043154061148e-05,
+      "loss": 0.9887,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597249498011906e-05,
+      "loss": 1.0426,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594064115014528e-05,
+      "loss": 0.9969,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590875392857008e-05,
+      "loss": 0.9794,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587683332778305e-05,
+      "loss": 0.9782,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584487936018663e-05,
+      "loss": 1.008,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858128920381963e-05,
+      "loss": 0.9237,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578087137424057e-05,
+      "loss": 1.0707,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574881738076076e-05,
+      "loss": 1.0352,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571673007021124e-05,
+      "loss": 0.9502,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856846094550593e-05,
+      "loss": 0.943,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565245554778516e-05,
+      "loss": 1.0207,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562026836088198e-05,
+      "loss": 0.9793,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855880479068559e-05,
+      "loss": 0.988,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855557941982258e-05,
+      "loss": 0.9818,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552350724752383e-05,
+      "loss": 1.0545,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854911870672947e-05,
+      "loss": 0.9812,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854588336700962e-05,
+      "loss": 0.9974,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85426447068499e-05,
+      "loss": 1.0514,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853940272750867e-05,
+      "loss": 1.048,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536157430245577e-05,
+      "loss": 0.9666,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532908816321557e-05,
+      "loss": 0.9721,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529656886998838e-05,
+      "loss": 0.9804,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526401643540924e-05,
+      "loss": 1.0311,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523143087212623e-05,
+      "loss": 1.0509,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851988121928002e-05,
+      "loss": 0.9502,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516616041010495e-05,
+      "loss": 0.9811,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85133475536727e-05,
+      "loss": 1.048,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851007575853659e-05,
+      "loss": 0.9852,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506800656873397e-05,
+      "loss": 0.9478,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850352224995563e-05,
+      "loss": 0.9916,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500240539057093e-05,
+      "loss": 1.0498,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496955525452877e-05,
+      "loss": 1.0343,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493667210419337e-05,
+      "loss": 0.998,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490375595234135e-05,
+      "loss": 0.9857,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487080681176195e-05,
+      "loss": 1.0434,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848378246952574e-05,
+      "loss": 1.0055,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 0.9706,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477176158574533e-05,
+      "loss": 0.9489,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473868061840617e-05,
+      "loss": 0.989,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847055667264785e-05,
+      "loss": 0.9174,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 1.0207,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463924022033492e-05,
+      "loss": 0.9662,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460602763188974e-05,
+      "loss": 0.9672,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457278217039735e-05,
+      "loss": 0.9508,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8453950384877504e-05,
+      "loss": 0.8781,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8450619267995283e-05,
+      "loss": 1.0562,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.844728486768736e-05,
+      "loss": 0.9251,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8443947185249287e-05,
+      "loss": 0.9532,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8440606221977893e-05,
+      "loss": 1.0463,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8437261979171294e-05,
+      "loss": 1.0444,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843391445812886e-05,
+      "loss": 0.9614,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843056366015125e-05,
+      "loss": 0.992,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427209586540392e-05,
+      "loss": 0.984,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423852238599488e-05,
+      "loss": 0.9614,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842049161763301e-05,
+      "loss": 1.0166,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84171277249467e-05,
+      "loss": 1.0409,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841376056184758e-05,
+      "loss": 1.0291,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410390129643927e-05,
+      "loss": 0.9532,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407016429645305e-05,
+      "loss": 1.0474,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403639463162536e-05,
+      "loss": 1.0346,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400259231507716e-05,
+      "loss": 1.0093,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839687573599421e-05,
+      "loss": 0.9844,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393488977936654e-05,
+      "loss": 0.9534,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839009895865094e-05,
+      "loss": 0.9561,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386705679454243e-05,
+      "loss": 0.9795,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383309141664992e-05,
+      "loss": 0.9441,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837990934660288e-05,
+      "loss": 1.0305,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376506295588887e-05,
+      "loss": 1.0152,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373099989945236e-05,
+      "loss": 0.9609,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836969043099542e-05,
+      "loss": 1.0065,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83662776200642e-05,
+      "loss": 1.0218,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362861558477597e-05,
+      "loss": 1.0548,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359442247562896e-05,
+      "loss": 0.9478,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835601968864865e-05,
+      "loss": 1.0399,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352593883064664e-05,
+      "loss": 1.021,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349164832142015e-05,
+      "loss": 1.0301,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834573253721303e-05,
+      "loss": 1.0136,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83422969996113e-05,
+      "loss": 0.9763,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338858220671683e-05,
+      "loss": 1.0031,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833541620173029e-05,
+      "loss": 1.0097,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833197094412449e-05,
+      "loss": 1.0125,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328522449192915e-05,
+      "loss": 0.9995,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325070718275452e-05,
+      "loss": 1.0953,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321615752713242e-05,
+      "loss": 1.0285,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318157553848694e-05,
+      "loss": 1.0101,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 1.0113,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311231461588446e-05,
+      "loss": 1.0515,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830776357088383e-05,
+      "loss": 1.0287,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304292452259037e-05,
+      "loss": 0.957,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300818107062744e-05,
+      "loss": 1.0373,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297340536644877e-05,
+      "loss": 0.9073,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293859742356625e-05,
+      "loss": 1.0365,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290375725550417e-05,
+      "loss": 1.0222,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286888487579954e-05,
+      "loss": 0.947,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283398029800167e-05,
+      "loss": 0.9453,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827990435356725e-05,
+      "loss": 1.0411,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.827640746023865e-05,
+      "loss": 0.9842,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.827290735117305e-05,
+      "loss": 0.9984,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82694040277304e-05,
+      "loss": 1.0187,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8265897491271885e-05,
+      "loss": 1.0273,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 1.0231,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258874784758276e-05,
+      "loss": 1.0165,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82553586174318e-05,
+      "loss": 0.9972,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251839242546696e-05,
+      "loss": 0.9825,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82483166614704e-05,
+      "loss": 0.9572,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244790875571582e-05,
+      "loss": 1.029,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241261886220155e-05,
+      "loss": 0.958,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237729694787295e-05,
+      "loss": 1.0415,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234194302645393e-05,
+      "loss": 0.9946,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230655711168112e-05,
+      "loss": 0.947,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227113921730336e-05,
+      "loss": 0.9656,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822356893570821e-05,
+      "loss": 1.0362,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220020754479104e-05,
+      "loss": 1.0073,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821646937942164e-05,
+      "loss": 0.9918,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821291481191568e-05,
+      "loss": 1.0137,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209357053342325e-05,
+      "loss": 1.0226,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205796105083917e-05,
+      "loss": 0.9983,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202231968524033e-05,
+      "loss": 1.0328,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198664645047496e-05,
+      "loss": 0.9421,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195094136040357e-05,
+      "loss": 1.0071,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819152044288992e-05,
+      "loss": 1.0017,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818794356698471e-05,
+      "loss": 1.0261,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184363509714494e-05,
+      "loss": 1.0146,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818078027247028e-05,
+      "loss": 1.0354,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177193856644315e-05,
+      "loss": 0.9661,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173604263630066e-05,
+      "loss": 0.9532,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170011494822242e-05,
+      "loss": 0.9995,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816641555161679e-05,
+      "loss": 1.0391,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162816435410892e-05,
+      "loss": 1.0135,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159214147602947e-05,
+      "loss": 0.921,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155608689592604e-05,
+      "loss": 0.9913,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152000062780735e-05,
+      "loss": 1.0182,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148388268569453e-05,
+      "loss": 1.0139,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814477330836208e-05,
+      "loss": 1.0692,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141155183563195e-05,
+      "loss": 0.9133,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137533895578585e-05,
+      "loss": 1.0725,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133909445815277e-05,
+      "loss": 1.0432,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813028183568153e-05,
+      "loss": 0.9211,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126651066586818e-05,
+      "loss": 1.004,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123017139941853e-05,
+      "loss": 0.992,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811938005715857e-05,
+      "loss": 1.0093,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115739819650124e-05,
+      "loss": 1.0237,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112096428830912e-05,
+      "loss": 0.9229,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108449886116543e-05,
+      "loss": 0.9865,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104800192923856e-05,
+      "loss": 0.9721,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101147350670905e-05,
+      "loss": 1.042,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097491360776987e-05,
+      "loss": 0.9984,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093832224662598e-05,
+      "loss": 0.9836,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.9906,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.808650451946057e-05,
+      "loss": 1.0409,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8082835953220055e-05,
+      "loss": 1.075,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8079164246453322e-05,
+      "loss": 1.0138,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8075489400586993e-05,
+      "loss": 0.9744,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807181141704889e-05,
+      "loss": 0.9933,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806813029726808e-05,
+      "loss": 0.9986,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806444604267483e-05,
+      "loss": 0.9381,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060758654700622e-05,
+      "loss": 1.0708,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057068134778174e-05,
+      "loss": 0.9881,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053374484341403e-05,
+      "loss": 1.0454,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049677704825455e-05,
+      "loss": 1.0152,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045977797666685e-05,
+      "loss": 1.0463,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042274764302664e-05,
+      "loss": 0.9158,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038568606172172e-05,
+      "loss": 0.9781,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803485932471522e-05,
+      "loss": 1.0215,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803114692137302e-05,
+      "loss": 1.0187,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027431397587993e-05,
+      "loss": 0.993,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023712754803783e-05,
+      "loss": 0.9706,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019990994465235e-05,
+      "loss": 0.9942,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801626611801842e-05,
+      "loss": 1.0399,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801253812691061e-05,
+      "loss": 0.955,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008807022590283e-05,
+      "loss": 0.9929,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005072806507132e-05,
+      "loss": 0.9902,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001335480112067e-05,
+      "loss": 1.0587,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799759504485719e-05,
+      "loss": 0.994,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993851502195822e-05,
+      "loss": 0.9729,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990104853582494e-05,
+      "loss": 0.9673,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798635510047293e-05,
+      "loss": 1.0479,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982602244324074e-05,
+      "loss": 0.9967,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978846286594073e-05,
+      "loss": 1.047,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975087228742272e-05,
+      "loss": 0.9218,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971325072229227e-05,
+      "loss": 0.9284,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967559818516697e-05,
+      "loss": 0.9895,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963791469067637e-05,
+      "loss": 1.0033,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796002002534622e-05,
+      "loss": 1.0676,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795624548881781e-05,
+      "loss": 1.0139,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952467860948975e-05,
+      "loss": 1.0378,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948687143207485e-05,
+      "loss": 1.0118,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944903337062304e-05,
+      "loss": 1.002,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941116443983612e-05,
+      "loss": 0.9121,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793732646544277e-05,
+      "loss": 1.017,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933533402912354e-05,
+      "loss": 0.9185,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929737257866123e-05,
+      "loss": 0.9419,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925938031779044e-05,
+      "loss": 1.0027,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922135726127276e-05,
+      "loss": 1.0133,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918330342388183e-05,
+      "loss": 1.0683,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791452188204031e-05,
+      "loss": 1.0395,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910710346563417e-05,
+      "loss": 1.0596,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906895737438437e-05,
+      "loss": 0.9788,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903078056147515e-05,
+      "loss": 0.9406,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899257304173975e-05,
+      "loss": 0.9875,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7895433483002356e-05,
+      "loss": 1.058,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7891606594118362e-05,
+      "loss": 0.9788,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7887776639008912e-05,
+      "loss": 1.0376,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7883943619162102e-05,
+      "loss": 1.0771,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.788010753606722e-05,
+      "loss": 1.0157,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876268391214756e-05,
+      "loss": 1.012,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872426186096374e-05,
+      "loss": 0.9045,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786858092220494e-05,
+      "loss": 1.0272,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78647326010345e-05,
+      "loss": 1.0354,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786088122408029e-05,
+      "loss": 0.9762,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785702679283874e-05,
+      "loss": 0.9999,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785316930880745e-05,
+      "loss": 0.9329,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849308773485226e-05,
+      "loss": 1.0061,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845445188372047e-05,
+      "loss": 0.9609,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784157855496908e-05,
+      "loss": 1.0131,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837708874778683e-05,
+      "loss": 0.9493,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783383614930438e-05,
+      "loss": 0.9863,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78299603800509e-05,
+      "loss": 1.0291,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782608156852414e-05,
+      "loss": 0.9641,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822199716231186e-05,
+      "loss": 0.9475,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 1.0326,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781442689538093e-05,
+      "loss": 0.9391,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810535929843703e-05,
+      "loss": 1.0212,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780664192958042e-05,
+      "loss": 0.9812,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780274489610407e-05,
+      "loss": 0.9481,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798844830928818e-05,
+      "loss": 1.0038,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794941735570005e-05,
+      "loss": 0.9961,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791035611544146e-05,
+      "loss": 0.9567,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778712646036894e-05,
+      "loss": 0.9099,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778321428356326e-05,
+      "loss": 0.9903,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777929908264715e-05,
+      "loss": 1.0059,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775380859141837e-05,
+      "loss": 0.9995,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777145961456971e-05,
+      "loss": 1.0133,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767535350454347e-05,
+      "loss": 0.9979,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763608068320487e-05,
+      "loss": 1.0252,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775967776969405e-05,
+      "loss": 0.9803,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755744456102123e-05,
+      "loss": 1.0584,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751808129072968e-05,
+      "loss": 1.0666,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747868790136012e-05,
+      "loss": 1.0145,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743926440821857e-05,
+      "loss": 0.9781,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739981082662275e-05,
+      "loss": 1.037,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736032717190208e-05,
+      "loss": 1.0109,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773208134593976e-05,
+      "loss": 0.9016,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772812697044621e-05,
+      "loss": 0.981,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724169592245996e-05,
+      "loss": 1.0673,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772020921287674e-05,
+      "loss": 0.986,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716245833877202e-05,
+      "loss": 1.0044,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712279456787334e-05,
+      "loss": 1.001,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708310083148242e-05,
+      "loss": 0.9661,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704337714502195e-05,
+      "loss": 0.9931,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700362352392632e-05,
+      "loss": 0.9482,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769638399836414e-05,
+      "loss": 1.0091,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.769240265396249e-05,
+      "loss": 1.0418,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7688418320734596e-05,
+      "loss": 0.9671,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7684431000228548e-05,
+      "loss": 1.0102,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7680440693993586e-05,
+      "loss": 0.9673,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7676447403580114e-05,
+      "loss": 1.0113,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672451130539697e-05,
+      "loss": 1.0059,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668451876425056e-05,
+      "loss": 0.97,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766444964279007e-05,
+      "loss": 0.9938,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.9901,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656436243180385e-05,
+      "loss": 0.981,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765242508031923e-05,
+      "loss": 0.963,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648410944164823e-05,
+      "loss": 1.008,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644393836276832e-05,
+      "loss": 0.9759,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640373758216075e-05,
+      "loss": 1.0268,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636350711544518e-05,
+      "loss": 0.9804,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632324697825288e-05,
+      "loss": 1.0175,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628295718622666e-05,
+      "loss": 1.0343,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762426377550208e-05,
+      "loss": 1.0518,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762022887003011e-05,
+      "loss": 1.0223,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761619100377449e-05,
+      "loss": 0.9495,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612150178304102e-05,
+      "loss": 1.0107,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608106395188983e-05,
+      "loss": 1.0059,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604059656000313e-05,
+      "loss": 0.9675,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600009962310417e-05,
+      "loss": 0.9515,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595957315692782e-05,
+      "loss": 0.9606,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759190171772203e-05,
+      "loss": 1.0241,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587843169973934e-05,
+      "loss": 0.9824,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583781674025415e-05,
+      "loss": 0.9007,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757971723145453e-05,
+      "loss": 1.001,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75756498438405e-05,
+      "loss": 0.9725,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571579512763665e-05,
+      "loss": 1.0018,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756750623980553e-05,
+      "loss": 1.0531,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563430026548737e-05,
+      "loss": 1.0199,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559350874577066e-05,
+      "loss": 0.9364,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555268785475438e-05,
+      "loss": 0.9889,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755118376082992e-05,
+      "loss": 1.1008,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547095802227723e-05,
+      "loss": 0.9572,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543004911257186e-05,
+      "loss": 1.0271,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75389110895078e-05,
+      "loss": 1.0012,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534814338570187e-05,
+      "loss": 0.8914,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530714660036112e-05,
+      "loss": 0.9392,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526612055498468e-05,
+      "loss": 0.9852,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522506526551303e-05,
+      "loss": 0.9822,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518398074789776e-05,
+      "loss": 0.9788,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514286701810203e-05,
+      "loss": 0.9724,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751017240921003e-05,
+      "loss": 0.9919,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506055198587835e-05,
+      "loss": 0.9155,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501935071543325e-05,
+      "loss": 1.0603,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497812029677344e-05,
+      "loss": 0.9461,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493686074591873e-05,
+      "loss": 0.9129,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489557207890025e-05,
+      "loss": 0.9956,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7485425431176032e-05,
+      "loss": 1.053,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.748129074605527e-05,
+      "loss": 0.9991,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7477153154134244e-05,
+      "loss": 1.015,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7473012657020577e-05,
+      "loss": 0.9619,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7468869256323036e-05,
+      "loss": 0.9693,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7464722953651504e-05,
+      "loss": 1.0228,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460573750616996e-05,
+      "loss": 0.9907,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456421648831658e-05,
+      "loss": 1.0231,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452266649908756e-05,
+      "loss": 1.0629,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448108755462684e-05,
+      "loss": 0.9379,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744394796710896e-05,
+      "loss": 1.0427,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439784286464234e-05,
+      "loss": 0.9845,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743561771514626e-05,
+      "loss": 1.0059,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 0.9595,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7427275906967288e-05,
+      "loss": 1.001,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423100673347435e-05,
+      "loss": 0.9492,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741892255553664e-05,
+      "loss": 0.9858,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741474155515827e-05,
+      "loss": 0.958,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741055767383683e-05,
+      "loss": 0.9897,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406370913197938e-05,
+      "loss": 0.992,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402181274868323e-05,
+      "loss": 0.9554,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397988760475842e-05,
+      "loss": 0.9859,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739379337164946e-05,
+      "loss": 0.9632,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738959511001928e-05,
+      "loss": 0.955,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385393977216486e-05,
+      "loss": 0.9971,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738118997487341e-05,
+      "loss": 1.038,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376983104623477e-05,
+      "loss": 1.0281,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737277336810124e-05,
+      "loss": 1.0568,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368560766942362e-05,
+      "loss": 0.9763,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736434530278362e-05,
+      "loss": 0.9453,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73601269772629e-05,
+      "loss": 1.0122,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355905792019195e-05,
+      "loss": 0.9041,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351681748692622e-05,
+      "loss": 0.9078,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73474548489244e-05,
+      "loss": 0.9544,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343225094356857e-05,
+      "loss": 1.0204,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338992486633438e-05,
+      "loss": 0.9716,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733475702739869e-05,
+      "loss": 0.9739,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 1.022,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732627756097893e-05,
+      "loss": 1.0173,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732203355708856e-05,
+      "loss": 0.9159,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317786708276124e-05,
+      "loss": 0.9846,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313537016191706e-05,
+      "loss": 1.0513,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309284482486494e-05,
+      "loss": 1.038,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305029108812777e-05,
+      "loss": 0.9947,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730077089682395e-05,
+      "loss": 1.0427,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729650984817451e-05,
+      "loss": 0.9872,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292245964520056e-05,
+      "loss": 0.9442,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287979247517285e-05,
+      "loss": 0.9732,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283709698824004e-05,
+      "loss": 0.9454,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727943732009911e-05,
+      "loss": 0.9792,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275162113002607e-05,
+      "loss": 1.0223,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270884079195596e-05,
+      "loss": 1.034,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7266603220340273e-05,
+      "loss": 0.9475,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.726231953809993e-05,
+      "loss": 0.9347,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7258033034138972e-05,
+      "loss": 0.9692,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7253743710122877e-05,
+      "loss": 1.0728,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7249451567718236e-05,
+      "loss": 0.9858,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7245156608592727e-05,
+      "loss": 0.988,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240858834415126e-05,
+      "loss": 1.031,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236558246855303e-05,
+      "loss": 0.9969,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232254847584212e-05,
+      "loss": 0.8557,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227948638273918e-05,
+      "loss": 1.0248,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223639620597556e-05,
+      "loss": 1.0168,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721932779622937e-05,
+      "loss": 0.9851,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721501316684469e-05,
+      "loss": 0.977,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210695734119926e-05,
+      "loss": 1.0493,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206375499732586e-05,
+      "loss": 1.0441,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202052465361268e-05,
+      "loss": 0.9964,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197726632685653e-05,
+      "loss": 1.0071,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193398003386514e-05,
+      "loss": 1.0154,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189066579145704e-05,
+      "loss": 0.9895,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184732361646168e-05,
+      "loss": 1.0068,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718039535257194e-05,
+      "loss": 0.9544,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717605555360812e-05,
+      "loss": 1.0543,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171712966440916e-05,
+      "loss": 1.029,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71673675927576e-05,
+      "loss": 0.9685,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163019434246545e-05,
+      "loss": 0.9704,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158668492597186e-05,
+      "loss": 1.0292,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154314769500052e-05,
+      "loss": 0.9376,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149958266646756e-05,
+      "loss": 0.9624,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145598985729976e-05,
+      "loss": 0.9444,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141236928443482e-05,
+      "loss": 1.0572,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136872096482123e-05,
+      "loss": 1.0113,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713250449154182e-05,
+      "loss": 1.0294,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128134115319567e-05,
+      "loss": 1.0029,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712376096951345e-05,
+      "loss": 1.0359,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711938505582262e-05,
+      "loss": 0.9697,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115006375947304e-05,
+      "loss": 1.0178,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110624931588805e-05,
+      "loss": 0.9507,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106240724449507e-05,
+      "loss": 0.9697,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101853756232856e-05,
+      "loss": 0.9359,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097464028643376e-05,
+      "loss": 1.0386,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7093071543386667e-05,
+      "loss": 0.9691,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088676302169394e-05,
+      "loss": 0.9922,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084278306699295e-05,
+      "loss": 0.967,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079877558685178e-05,
+      "loss": 1.0359,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075474059836923e-05,
+      "loss": 1.0272,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.9961,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066658816482856e-05,
+      "loss": 1.1309,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7062247075402137e-05,
+      "loss": 1.0224,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705783259033748e-05,
+      "loss": 0.9915,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705341536300409e-05,
+      "loss": 1.0015,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048995395118253e-05,
+      "loss": 0.9788,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044572688397316e-05,
+      "loss": 0.9605,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7040147244559688e-05,
+      "loss": 0.9653,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7035719065324837e-05,
+      "loss": 0.9983,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7031288152413304e-05,
+      "loss": 1.0337,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7026854507546694e-05,
+      "loss": 0.9763,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7022418132447656e-05,
+      "loss": 0.9769,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7017979028839918e-05,
+      "loss": 1.061,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013537198448253e-05,
+      "loss": 1.0048,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700909264299851e-05,
+      "loss": 1.0526,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004645364217584e-05,
+      "loss": 0.9605,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000195363833434e-05,
+      "loss": 1.0058,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6995742643575076e-05,
+      "loss": 0.969,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991287205172575e-05,
+      "loss": 1.0376,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6986829050357068e-05,
+      "loss": 0.928,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698236818086073e-05,
+      "loss": 1.0356,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977904598416803e-05,
+      "loss": 0.9948,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697343830475958e-05,
+      "loss": 1.0623,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968969301624402e-05,
+      "loss": 0.9727,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696449759074767e-05,
+      "loss": 0.9647,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960023173866834e-05,
+      "loss": 0.9351,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955546052720398e-05,
+      "loss": 0.9832,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695106622904791e-05,
+      "loss": 0.9709,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946583704589973e-05,
+      "loss": 0.9398,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942098481088245e-05,
+      "loss": 1.0096,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693761056028542e-05,
+      "loss": 0.9211,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6933119943925243e-05,
+      "loss": 0.9244,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692862663375252e-05,
+      "loss": 1.0646,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692413063151309e-05,
+      "loss": 0.9418,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919631938953835e-05,
+      "loss": 0.9885,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915130557822698e-05,
+      "loss": 1.0436,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691062648986865e-05,
+      "loss": 0.9684,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690611973684172e-05,
+      "loss": 1.025,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901610300492963e-05,
+      "loss": 0.9638,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897098182574502e-05,
+      "loss": 0.9739,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689258338483947e-05,
+      "loss": 0.9821,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888065909042074e-05,
+      "loss": 1.0658,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688354575693754e-05,
+      "loss": 0.9168,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687902293028214e-05,
+      "loss": 0.9943,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874497430833182e-05,
+      "loss": 0.9808,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686996926034902e-05,
+      "loss": 0.9873,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865438420589036e-05,
+      "loss": 0.9689,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6860904913313656e-05,
+      "loss": 1.0239,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6856368740284342e-05,
+      "loss": 1.0142,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685182990326359e-05,
+      "loss": 1.0035,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847288404014937e-05,
+      "loss": 0.9669,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684274424430294e-05,
+      "loss": 1.024,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68381974258932e-05,
+      "loss": 0.9855,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833647950552357e-05,
+      "loss": 0.9986,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682909582004807e-05,
+      "loss": 0.9358,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682454103614904e-05,
+      "loss": 0.961,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819983600624986e-05,
+      "loss": 0.9362,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6815423515246674e-05,
+      "loss": 1.0259,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681086078178589e-05,
+      "loss": 1.0158,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.680629540201545e-05,
+      "loss": 0.9745,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6801727377709195e-05,
+      "loss": 1.0546,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6797156710642003e-05,
+      "loss": 1.0001,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.679258340258977e-05,
+      "loss": 1.1047,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.678800745532942e-05,
+      "loss": 0.9375,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6783428870638904e-05,
+      "loss": 0.9767,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67788476502972e-05,
+      "loss": 1.022,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774263796084304e-05,
+      "loss": 0.9351,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676967730978124e-05,
+      "loss": 1.0144,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765088193170055e-05,
+      "loss": 1.0672,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676049644803381e-05,
+      "loss": 1.0465,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755902076156606e-05,
+      "loss": 1.0073,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6751305079323538e-05,
+      "loss": 1.0335,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746705459320746e-05,
+      "loss": 0.9246,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742103217935372e-05,
+      "loss": 1.0008,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737498356955586e-05,
+      "loss": 1.018,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732890878170573e-05,
+      "loss": 0.9976,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672828078337053e-05,
+      "loss": 1.027,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672366807434668e-05,
+      "loss": 0.9972,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719052752891262e-05,
+      "loss": 0.966,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714434820797515e-05,
+      "loss": 1.026,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67098142798597e-05,
+      "loss": 0.9705,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670519113187311e-05,
+      "loss": 0.9853,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6700565378634022e-05,
+      "loss": 0.9263,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669593702193974e-05,
+      "loss": 1.0117,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.9555,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668667250537987e-05,
+      "loss": 0.9695,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668203634911394e-05,
+      "loss": 1.0401,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677397596592134e-05,
+      "loss": 0.9769,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667275624961681e-05,
+      "loss": 1.0202,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666811230999132e-05,
+      "loss": 1.0427,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663465779520042e-05,
+      "loss": 1.0111,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658816660008344e-05,
+      "loss": 1.0367,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6654164953262614e-05,
+      "loss": 1.0063,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649510661090233e-05,
+      "loss": 0.9566,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644853785299592e-05,
+      "loss": 0.9062,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640194327700087e-05,
+      "loss": 0.9555,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635532290102114e-05,
+      "loss": 1.0019,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630867674317078e-05,
+      "loss": 0.9846,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626200482157378e-05,
+      "loss": 0.9422,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621530715436413e-05,
+      "loss": 1.0374,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616858375968596e-05,
+      "loss": 1.0757,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612183465569322e-05,
+      "loss": 0.9875,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607505986054998e-05,
+      "loss": 0.985,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660282593924302e-05,
+      "loss": 0.9834,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598143326951784e-05,
+      "loss": 0.9725,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659345815100069e-05,
+      "loss": 0.9395,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588770413210126e-05,
+      "loss": 0.993,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584080115401478e-05,
+      "loss": 0.9238,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657938725939713e-05,
+      "loss": 0.9466,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574691847020446e-05,
+      "loss": 1.0173,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569993880095807e-05,
+      "loss": 0.9725,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6565293360448565e-05,
+      "loss": 1.0072,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6560590289905074e-05,
+      "loss": 1.0424,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.655588467029268e-05,
+      "loss": 1.0255,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6551176503439713e-05,
+      "loss": 1.065,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6546465791175498e-05,
+      "loss": 0.9988,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6541752535330345e-05,
+      "loss": 1.041,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537036737735563e-05,
+      "loss": 1.0259,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532318400223427e-05,
+      "loss": 0.9812,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527597524627226e-05,
+      "loss": 1.0523,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522874112781213e-05,
+      "loss": 1.0216,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6518148166520638e-05,
+      "loss": 0.9501,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651341968768173e-05,
+      "loss": 1.0609,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508688678101706e-05,
+      "loss": 0.9897,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503955139618765e-05,
+      "loss": 0.9359,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499219074072087e-05,
+      "loss": 0.9971,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494480483301836e-05,
+      "loss": 1.0332,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648973936914916e-05,
+      "loss": 0.9797,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484995733456178e-05,
+      "loss": 0.9391,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480249578066e-05,
+      "loss": 0.943,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475500904822707e-05,
+      "loss": 0.9998,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647074971557136e-05,
+      "loss": 0.9485,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465996012157996e-05,
+      "loss": 0.9999,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646123979642964e-05,
+      "loss": 0.9375,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456481070234277e-05,
+      "loss": 0.9837,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645171983542088e-05,
+      "loss": 1.0266,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446956093839385e-05,
+      "loss": 1.0421,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644218984734071e-05,
+      "loss": 1.0013,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437421097776754e-05,
+      "loss": 1.0295,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643264984700037e-05,
+      "loss": 1.0076,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.972,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6423099849226633e-05,
+      "loss": 0.9736,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418321105939863e-05,
+      "loss": 0.9717,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6413539868861828e-05,
+      "loss": 0.955,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6408756139850243e-05,
+      "loss": 0.9526,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640396992076379e-05,
+      "loss": 1.0541,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639918121346212e-05,
+      "loss": 1.0016,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639439001980585e-05,
+      "loss": 0.9103,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638959634165656e-05,
+      "loss": 1.039,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384800180876804e-05,
+      "loss": 0.9889,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6380001539330088e-05,
+      "loss": 0.9751,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375200418880895e-05,
+      "loss": 1.0116,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637039682139466e-05,
+      "loss": 0.9908,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365590748737786e-05,
+      "loss": 0.9692,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636078220277764e-05,
+      "loss": 1.0697,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355971185382547e-05,
+      "loss": 1.0722,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635115769842179e-05,
+      "loss": 1.0264,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6346341743765616e-05,
+      "loss": 1.023,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634152332328522e-05,
+      "loss": 0.9245,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633670243885278e-05,
+      "loss": 0.9933,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331879092341402e-05,
+      "loss": 1.0479,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6327053285625164e-05,
+      "loss": 0.9932,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63222250205791e-05,
+      "loss": 1.0525,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6317394299079193e-05,
+      "loss": 1.031,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.631256112300239e-05,
+      "loss": 1.0062,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6307725494226586e-05,
+      "loss": 1.0236,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.630288741463062e-05,
+      "loss": 0.9892,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62980468860943e-05,
+      "loss": 0.9991,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6293203910498375e-05,
+      "loss": 0.9852,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628835848972455e-05,
+      "loss": 0.9585,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283510625655474e-05,
+      "loss": 0.9246,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627866032017475e-05,
+      "loss": 0.9728,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273807575166927e-05,
+      "loss": 0.9617,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268952392517505e-05,
+      "loss": 1.0079,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626409477411293e-05,
+      "loss": 0.9885,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259234721840595e-05,
+      "loss": 0.9667,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625437223758883e-05,
+      "loss": 0.9774,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249507323246926e-05,
+      "loss": 1.0301,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244639980705104e-05,
+      "loss": 0.932,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623977021185454e-05,
+      "loss": 0.9543,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 1.0233,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623002340279656e-05,
+      "loss": 1.0131,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225146366376198e-05,
+      "loss": 0.9857,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622026691122119e-05,
+      "loss": 1.0065,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621538503922741e-05,
+      "loss": 0.9254,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210500752291682e-05,
+      "loss": 0.9013,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620561405231175e-05,
+      "loss": 1.0213,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200724941186305e-05,
+      "loss": 1.0069,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195833420814983e-05,
+      "loss": 0.9559,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190939493098344e-05,
+      "loss": 1.0066,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186043159937884e-05,
+      "loss": 1.0571,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6181144423236038e-05,
+      "loss": 1.0172,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617624328489618e-05,
+      "loss": 1.0641,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6171339746822606e-05,
+      "loss": 0.9329,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616643381092055e-05,
+      "loss": 0.973,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616152547909618e-05,
+      "loss": 1.006,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156614753256583e-05,
+      "loss": 0.9597,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151701635309797e-05,
+      "loss": 0.9986,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146786127164773e-05,
+      "loss": 1.0326,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141868230731393e-05,
+      "loss": 0.9845,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6136947947920477e-05,
+      "loss": 0.9689,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132025280643753e-05,
+      "loss": 0.967,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6127100230813897e-05,
+      "loss": 0.899,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122172800344494e-05,
+      "loss": 0.9444,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117242991150064e-05,
+      "loss": 1.006,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611231080514605e-05,
+      "loss": 0.9532,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610737624424881e-05,
+      "loss": 0.9965,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102439310375637e-05,
+      "loss": 0.8703,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609750000544474e-05,
+      "loss": 1.011,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092558331375245e-05,
+      "loss": 0.9743,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608761429008721e-05,
+      "loss": 1.0183,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082667883501596e-05,
+      "loss": 0.9874,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077719113540303e-05,
+      "loss": 0.958,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607276798212613e-05,
+      "loss": 0.9954,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067814491182806e-05,
+      "loss": 0.9567,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.606285864263498e-05,
+      "loss": 1.0394,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60579004384082e-05,
+      "loss": 0.9835,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6052939880428942e-05,
+      "loss": 1.0061,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6047976970624603e-05,
+      "loss": 1.0013,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6043011710923476e-05,
+      "loss": 0.9715,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6038044103254775e-05,
+      "loss": 0.939,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603307414954863e-05,
+      "loss": 0.9606,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028101851736083e-05,
+      "loss": 1.0267,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023127211749082e-05,
+      "loss": 0.9691,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6018150231520486e-05,
+      "loss": 1.0173,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601317091298406e-05,
+      "loss": 1.0776,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6008189258074483e-05,
+      "loss": 0.991,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600320526872735e-05,
+      "loss": 1.0125,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599821894687914e-05,
+      "loss": 1.0247,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993230294467257e-05,
+      "loss": 1.0332,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988239313430004e-05,
+      "loss": 0.9404,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983246005706592e-05,
+      "loss": 1.002,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978250373237132e-05,
+      "loss": 0.9767,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597325241796264e-05,
+      "loss": 0.9043,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5968252141825038e-05,
+      "loss": 0.9674,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963249546767144e-05,
+      "loss": 0.9417,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958244634732673e-05,
+      "loss": 1.0007,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595323740766626e-05,
+      "loss": 0.9838,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948227867513416e-05,
+      "loss": 0.9634,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943216016220563e-05,
+      "loss": 0.9519,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938201855735017e-05,
+      "loss": 1.0721,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5933185388004998e-05,
+      "loss": 1.0831,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928166614979613e-05,
+      "loss": 0.9793,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923145538608874e-05,
+      "loss": 0.9161,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591812216084368e-05,
+      "loss": 1.0048,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5913096483635827e-05,
+      "loss": 1.0056,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908068508938e-05,
+      "loss": 1.0066,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903038238703795e-05,
+      "loss": 0.9517,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898005674887673e-05,
+      "loss": 0.9938,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589297081944501e-05,
+      "loss": 0.9588,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887933674332048e-05,
+      "loss": 0.8606,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882894241505946e-05,
+      "loss": 1.0097,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 1.0167,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5872808520547334e-05,
+      "loss": 1.0348,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5867762236333553e-05,
+      "loss": 0.906,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862713672244092e-05,
+      "loss": 0.9894,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585766283024053e-05,
+      "loss": 1.0097,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585260971228534e-05,
+      "loss": 0.947,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5847554320341865e-05,
+      "loss": 1.0037,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842496656374347e-05,
+      "loss": 0.9772,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837436722347902e-05,
+      "loss": 0.9611,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832374520228527e-05,
+      "loss": 0.9452,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582731005198311e-05,
+      "loss": 0.9323,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822243319579405e-05,
+      "loss": 0.9431,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581717432498606e-05,
+      "loss": 1.0014,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812103070172592e-05,
+      "loss": 0.9957,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807029557109398e-05,
+      "loss": 0.9836,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.580195378776776e-05,
+      "loss": 0.9524,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5796875764119826e-05,
+      "loss": 0.9701,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5791795488138623e-05,
+      "loss": 1.0748,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.578671296179806e-05,
+      "loss": 1.0192,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.578162818707291e-05,
+      "loss": 0.8944,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577654116593883e-05,
+      "loss": 0.98,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577145190037234e-05,
+      "loss": 0.9943,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766360392350837e-05,
+      "loss": 0.9545,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761266643852587e-05,
+      "loss": 0.9234,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575617065685674e-05,
+      "loss": 0.9848,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751072433343285e-05,
+      "loss": 0.9997,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745971975293114e-05,
+      "loss": 0.9854,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574086928468797e-05,
+      "loss": 0.9739,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573576436351046e-05,
+      "loss": 1.0348,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5730657213744072e-05,
+      "loss": 0.9505,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725547837373142e-05,
+      "loss": 1.0087,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720436236382885e-05,
+      "loss": 1.0731,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715322412759374e-05,
+      "loss": 0.9593,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5710206368489555e-05,
+      "loss": 0.9604,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705088105561217e-05,
+      "loss": 0.9738,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699967625963032e-05,
+      "loss": 0.9844,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569484493168452e-05,
+      "loss": 0.9504,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568972002471607e-05,
+      "loss": 0.9274,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684592907048925e-05,
+      "loss": 0.9471,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5679463580675187e-05,
+      "loss": 0.9659,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567433204758782e-05,
+      "loss": 0.9832,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5669198309780644e-05,
+      "loss": 1.0491,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566406236924833e-05,
+      "loss": 0.9998,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658924227986415e-05,
+      "loss": 1.0008,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653783887991282e-05,
+      "loss": 0.978,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564864135126018e-05,
+      "loss": 0.949,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564349661979119e-05,
+      "loss": 0.9464,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638349695583276e-05,
+      "loss": 0.9778,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563320058063622e-05,
+      "loss": 1.0406,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628049276950687e-05,
+      "loss": 1.038,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622895786528173e-05,
+      "loss": 0.9445,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617740111371023e-05,
+      "loss": 1.0716,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612582253482444e-05,
+      "loss": 1.0109,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560742221486648e-05,
+      "loss": 0.946,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602259997528028e-05,
+      "loss": 0.9948,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597095603472828e-05,
+      "loss": 1.0127,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 1.0066,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586760293239383e-05,
+      "loss": 0.9076,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5581589381076843e-05,
+      "loss": 1.0529,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557641630022897e-05,
+      "loss": 1.0398,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571241052705724e-05,
+      "loss": 0.9034,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566063640517918e-05,
+      "loss": 1.0109,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560884065677188e-05,
+      "loss": 1.0128,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.9617,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550518436087753e-05,
+      "loss": 1.0221,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554533238536653e-05,
+      "loss": 0.9328,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540144180047367e-05,
+      "loss": 1.0066,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5534953822146097e-05,
+      "loss": 1.0095,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5529761313679396e-05,
+      "loss": 0.9862,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5524566656664777e-05,
+      "loss": 0.9735,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5519369853120584e-05,
+      "loss": 1.0077,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5514170905065996e-05,
+      "loss": 0.9877,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5508969814521026e-05,
+      "loss": 0.9635,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5503766583506522e-05,
+      "loss": 0.9763,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549856121404416e-05,
+      "loss": 0.9622,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493353708156442e-05,
+      "loss": 0.9922,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548814406786671e-05,
+      "loss": 1.0025,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548293229519914e-05,
+      "loss": 1.0052,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477718392178716e-05,
+      "loss": 0.9845,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472502360831265e-05,
+      "loss": 0.982,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467284203183437e-05,
+      "loss": 0.948,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462063921262708e-05,
+      "loss": 1.008,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456841517097385e-05,
+      "loss": 1.0074,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545161699271659e-05,
+      "loss": 0.9624,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446390350150272e-05,
+      "loss": 0.9337,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544116159142921e-05,
+      "loss": 1.0418,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435930718584993e-05,
+      "loss": 0.9547,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430697733650045e-05,
+      "loss": 0.9493,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425462638657597e-05,
+      "loss": 0.9304,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542022543564171e-05,
+      "loss": 0.8934,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541498612663726e-05,
+      "loss": 0.983,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5409744713679942e-05,
+      "loss": 0.9807,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404501198806267e-05,
+      "loss": 0.9807,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399255584053568e-05,
+      "loss": 1.0032,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394007871459983e-05,
+      "loss": 0.9782,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538875806306448e-05,
+      "loss": 0.9761,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383506160906826e-05,
+      "loss": 1.0376,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378252167027613e-05,
+      "loss": 1.012,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372996083468242e-05,
+      "loss": 0.9333,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536773791227092e-05,
+      "loss": 0.9387,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362477655478677e-05,
+      "loss": 0.9999,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535721531513535e-05,
+      "loss": 0.969,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535195089328557e-05,
+      "loss": 0.9497,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346684391974792e-05,
+      "loss": 0.9695,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534141581324929e-05,
+      "loss": 0.986,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533614515915612e-05,
+      "loss": 0.9232,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533087243174315e-05,
+      "loss": 0.9334,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325597633059075e-05,
+      "loss": 0.9661,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.9788,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315041830076317e-05,
+      "loss": 0.9331,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530976082987902e-05,
+      "loss": 0.9628,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304477766613365e-05,
+      "loss": 0.9699,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529919264233205e-05,
+      "loss": 1.0131,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529390545908857e-05,
+      "loss": 0.9756,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528861621893722e-05,
+      "loss": 0.9775,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52833249239331e-05,
+      "loss": 0.9462,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527803157613209e-05,
+      "loss": 1.0349,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272736177590902e-05,
+      "loss": 0.9536,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526743873036701e-05,
+      "loss": 1.0358,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5262139236518695e-05,
+      "loss": 0.9722,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5256837698105047e-05,
+      "loss": 0.966,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5251534117185935e-05,
+      "loss": 0.997,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5246228495822025e-05,
+      "loss": 0.9901,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5240920836074777e-05,
+      "loss": 1.0381,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5235611140006446e-05,
+      "loss": 0.9399,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230299409680073e-05,
+      "loss": 0.9582,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5224985647159489e-05,
+      "loss": 1.0111,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219669854509322e-05,
+      "loss": 1.0512,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214352033794981e-05,
+      "loss": 0.9755,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209032187082668e-05,
+      "loss": 0.9704,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203710316439368e-05,
+      "loss": 1.0135,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198386423932856e-05,
+      "loss": 0.9868,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193060511631692e-05,
+      "loss": 0.8768,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187732581605217e-05,
+      "loss": 1.0268,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518240263592356e-05,
+      "loss": 1.0544,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177070676657632e-05,
+      "loss": 0.8994,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171736705879127e-05,
+      "loss": 0.9358,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166400725660522e-05,
+      "loss": 0.9769,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161062738075068e-05,
+      "loss": 0.9559,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155722745196802e-05,
+      "loss": 0.9895,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150380749100545e-05,
+      "loss": 0.9282,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514503675186188e-05,
+      "loss": 0.913,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139690755557186e-05,
+      "loss": 0.9467,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134342762263606e-05,
+      "loss": 1.0192,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128992774059063e-05,
+      "loss": 1.0281,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123640793022255e-05,
+      "loss": 0.9853,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118286821232661e-05,
+      "loss": 0.984,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511293086077052e-05,
+      "loss": 0.9491,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107572913716859e-05,
+      "loss": 0.9858,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510221298215346e-05,
+      "loss": 0.9425,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096851068162885e-05,
+      "loss": 1.0141,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091487173828471e-05,
+      "loss": 1.0026,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086121301234318e-05,
+      "loss": 0.9392,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080753452465296e-05,
+      "loss": 0.8932,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075383629607043e-05,
+      "loss": 0.9764,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070011834745963e-05,
+      "loss": 0.9181,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064638069969228e-05,
+      "loss": 1.0179,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059262337364779e-05,
+      "loss": 0.9574,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505388463902131e-05,
+      "loss": 0.8839,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048504977028289e-05,
+      "loss": 1.0173,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043123353475944e-05,
+      "loss": 0.9988,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037739770455263e-05,
+      "loss": 0.9891,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032354230058004e-05,
+      "loss": 1.0046,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502696673437667e-05,
+      "loss": 0.9832,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021577285504538e-05,
+      "loss": 1.0069,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016185885535639e-05,
+      "loss": 1.0313,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010792536564762e-05,
+      "loss": 1.0054,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005397240687447e-05,
+      "loss": 0.9937,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9223,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994600816599482e-05,
+      "loss": 0.9798,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989199692583705e-05,
+      "loss": 0.9542,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.498379663005123e-05,
+      "loss": 0.9833,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4978391631101383e-05,
+      "loss": 0.9341,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4972984697834238e-05,
+      "loss": 0.9789,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4967575832350614e-05,
+      "loss": 1.0343,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4962165036752085e-05,
+      "loss": 0.9594,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4956752313140978e-05,
+      "loss": 0.9743,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951337663620368e-05,
+      "loss": 0.983,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945921090294076e-05,
+      "loss": 0.936,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494050259526667e-05,
+      "loss": 1.0174,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493508218064347e-05,
+      "loss": 0.9919,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929659848530537e-05,
+      "loss": 0.9697,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924235601034673e-05,
+      "loss": 0.9892,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918809440263435e-05,
+      "loss": 0.939,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913381368325115e-05,
+      "loss": 0.9918,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907951387328752e-05,
+      "loss": 0.946,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4902519499384122e-05,
+      "loss": 0.9346,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489708570660175e-05,
+      "loss": 1.0335,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891650011092896e-05,
+      "loss": 0.9348,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886212414969551e-05,
+      "loss": 0.9631,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488077292034446e-05,
+      "loss": 1.0051,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875331529331097e-05,
+      "loss": 0.9429,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869888244043674e-05,
+      "loss": 1.0105,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486444306659714e-05,
+      "loss": 0.9339,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858995999107175e-05,
+      "loss": 0.9383,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853547043690201e-05,
+      "loss": 0.9364,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848096202463373e-05,
+      "loss": 1.0337,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842643477544565e-05,
+      "loss": 0.9239,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837188871052399e-05,
+      "loss": 0.9728,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831732385106221e-05,
+      "loss": 0.9213,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482627402182611e-05,
+      "loss": 0.931,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482081378333287e-05,
+      "loss": 0.9794,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815351671748039e-05,
+      "loss": 0.9754,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809887689193878e-05,
+      "loss": 0.9522,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804421837793379e-05,
+      "loss": 0.942,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798954119670254e-05,
+      "loss": 0.974,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479348453694895e-05,
+      "loss": 0.9926,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4788013091754629e-05,
+      "loss": 0.9328,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782539786213184e-05,
+      "loss": 0.9619,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777064622451222e-05,
+      "loss": 0.9673,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771587602596085e-05,
+      "loss": 0.9988,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766108728775822e-05,
+      "loss": 0.997,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760628003119213e-05,
+      "loss": 1.0168,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755145427755755e-05,
+      "loss": 0.9547,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749661004815654e-05,
+      "loss": 0.9859,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474417473642985e-05,
+      "loss": 0.8581,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738686624729987e-05,
+      "loss": 0.9564,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733196671848435e-05,
+      "loss": 0.9641,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727704879918272e-05,
+      "loss": 1.0092,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722211251073293e-05,
+      "loss": 1.012,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716715787448007e-05,
+      "loss": 0.9507,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711218491177636e-05,
+      "loss": 0.9902,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705719364398117e-05,
+      "loss": 0.9344,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4700218409246087e-05,
+      "loss": 0.997,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.469471562785891e-05,
+      "loss": 1.0391,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4689211022374645e-05,
+      "loss": 1.0774,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4683704594932072e-05,
+      "loss": 0.9877,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4678196347670665e-05,
+      "loss": 0.9367,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4672686282730622e-05,
+      "loss": 0.9897,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4667174402252831e-05,
+      "loss": 0.9941,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661660708378893e-05,
+      "loss": 0.9732,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656145203251116e-05,
+      "loss": 0.9417,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650627889012507e-05,
+      "loss": 0.9461,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645108767806778e-05,
+      "loss": 1.0302,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639587841778342e-05,
+      "loss": 1.054,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634065113072316e-05,
+      "loss": 0.9997,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628540583834511e-05,
+      "loss": 0.9938,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462301425621145e-05,
+      "loss": 0.9568,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617486132350343e-05,
+      "loss": 0.9632,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611956214399097e-05,
+      "loss": 1.0235,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606424504506325e-05,
+      "loss": 0.9642,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600891004821332e-05,
+      "loss": 0.9891,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595355717494118e-05,
+      "loss": 1.0003,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589818644675378e-05,
+      "loss": 0.9786,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45842797885165e-05,
+      "loss": 1.0615,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578739151169567e-05,
+      "loss": 0.9336,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573196734787357e-05,
+      "loss": 1.0272,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567652541523325e-05,
+      "loss": 0.8923,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562106573531632e-05,
+      "loss": 0.9573,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556558832967122e-05,
+      "loss": 0.9933,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455100932198533e-05,
+      "loss": 1.0441,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545458042742477e-05,
+      "loss": 0.9636,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 0.991,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534350188101905e-05,
+      "loss": 1.0308,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528793617020063e-05,
+      "loss": 1.0042,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523235286308907e-05,
+      "loss": 1.0277,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517675198128086e-05,
+      "loss": 1.0035,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512113354637931e-05,
+      "loss": 0.8561,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506549757999456e-05,
+      "loss": 0.9862,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500984410374353e-05,
+      "loss": 0.9672,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495417313924996e-05,
+      "loss": 0.9675,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489848470814447e-05,
+      "loss": 0.9864,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484277883206433e-05,
+      "loss": 0.9989,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478705553265363e-05,
+      "loss": 0.963,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473131483156326e-05,
+      "loss": 0.9486,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467555675045088e-05,
+      "loss": 0.881,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4461978131098089e-05,
+      "loss": 0.9838,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456398853482438e-05,
+      "loss": 0.934,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450817844365924e-05,
+      "loss": 1.0086,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444523510591701e-05,
+      "loss": 0.9334,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439650640304822e-05,
+      "loss": 0.937,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434064449699168e-05,
+      "loss": 0.9699,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428476536270517e-05,
+      "loss": 1.0032,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422886902190014e-05,
+      "loss": 0.9842,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4417295549629469e-05,
+      "loss": 0.9798,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4411702480761362e-05,
+      "loss": 0.969,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4406107697758838e-05,
+      "loss": 0.9654,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.440051120279571e-05,
+      "loss": 0.9648,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4394912998046451e-05,
+      "loss": 0.9894,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4389313085686205e-05,
+      "loss": 0.9944,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4383711467890776e-05,
+      "loss": 1.0253,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378108146836632e-05,
+      "loss": 0.9871,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372503124700899e-05,
+      "loss": 1.0211,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436689640366137e-05,
+      "loss": 0.9388,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361287985896495e-05,
+      "loss": 0.9636,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4355677873585383e-05,
+      "loss": 0.9695,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43500660689078e-05,
+      "loss": 0.9171,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344452574044173e-05,
+      "loss": 0.9963,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 1.019,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333220522483768e-05,
+      "loss": 0.9643,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327601970151122e-05,
+      "loss": 1.0084,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432198173636069e-05,
+      "loss": 1.075,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316359823296174e-05,
+      "loss": 0.9935,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310736233141926e-05,
+      "loss": 1.044,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305110968082953e-05,
+      "loss": 1.0064,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299484030304906e-05,
+      "loss": 0.9741,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293855421994094e-05,
+      "loss": 0.9391,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288225145337467e-05,
+      "loss": 0.9607,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4282593202522627e-05,
+      "loss": 0.9607,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276959595737832e-05,
+      "loss": 0.9683,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271324327171969e-05,
+      "loss": 0.967,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265687399014584e-05,
+      "loss": 0.9451,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260048813455867e-05,
+      "loss": 0.9787,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254408572686642e-05,
+      "loss": 0.9602,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248766678898386e-05,
+      "loss": 0.881,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243123134283217e-05,
+      "loss": 0.9905,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237477941033888e-05,
+      "loss": 0.8776,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231831101343802e-05,
+      "loss": 0.9369,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226182617406996e-05,
+      "loss": 1.0592,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220532491418144e-05,
+      "loss": 0.9724,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214880725572563e-05,
+      "loss": 0.9625,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209227322066206e-05,
+      "loss": 1.0516,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203572283095657e-05,
+      "loss": 0.9749,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197915610858143e-05,
+      "loss": 1.0213,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419225730755152e-05,
+      "loss": 0.9146,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186597375374283e-05,
+      "loss": 0.9474,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4180935816525554e-05,
+      "loss": 0.9239,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417527263320509e-05,
+      "loss": 1.0029,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169607827613284e-05,
+      "loss": 0.9693,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163941401951147e-05,
+      "loss": 1.0267,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415827335842033e-05,
+      "loss": 0.8752,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152603699223112e-05,
+      "loss": 1.0167,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146932426562391e-05,
+      "loss": 1.021,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141259542641706e-05,
+      "loss": 0.9949,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135585049665207e-05,
+      "loss": 0.9616,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129908949837684e-05,
+      "loss": 0.9079,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124231245364534e-05,
+      "loss": 0.9812,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4118551938451797e-05,
+      "loss": 1.0286,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4112871031306118e-05,
+      "loss": 0.9943,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4107188526134774e-05,
+      "loss": 0.9655,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4101504425145662e-05,
+      "loss": 0.8651,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4095818730547297e-05,
+      "loss": 0.9658,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4090131444548814e-05,
+      "loss": 0.9805,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084442569359964e-05,
+      "loss": 0.8628,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078752107191118e-05,
+      "loss": 1.0027,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073060060253265e-05,
+      "loss": 0.8852,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.9752,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061671220917557e-05,
+      "loss": 0.971,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055974432944753e-05,
+      "loss": 0.9682,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050276069053036e-05,
+      "loss": 1.0454,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044576131456466e-05,
+      "loss": 0.9619,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038874622369714e-05,
+      "loss": 0.9932,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033171544008053e-05,
+      "loss": 0.9234,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027466898587375e-05,
+      "loss": 0.9443,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021760688324175e-05,
+      "loss": 1.0026,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016052915435565e-05,
+      "loss": 0.9145,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010343582139252e-05,
+      "loss": 0.962,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004632690653559e-05,
+      "loss": 0.9613,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998920243197408e-05,
+      "loss": 1.0059,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993206241990333e-05,
+      "loss": 0.9931,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987490689252463e-05,
+      "loss": 0.9151,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981773587204535e-05,
+      "loss": 0.9838,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976054938067885e-05,
+      "loss": 1.0239,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970334744064451e-05,
+      "loss": 0.9801,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3964613007416777e-05,
+      "loss": 0.9756,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958889730347997e-05,
+      "loss": 0.95,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953164915081852e-05,
+      "loss": 1.0068,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947438563842672e-05,
+      "loss": 0.9224,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3941710678855396e-05,
+      "loss": 1.0093,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935981262345541e-05,
+      "loss": 0.9578,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930250316539237e-05,
+      "loss": 0.9223,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39245178436632e-05,
+      "loss": 0.9265,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3918783845944737e-05,
+      "loss": 1.0597,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391304832561175e-05,
+      "loss": 0.9625,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907311284892737e-05,
+      "loss": 0.9949,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901572726016782e-05,
+      "loss": 0.9426,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389583265121356e-05,
+      "loss": 0.9295,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890091062713332e-05,
+      "loss": 1.0159,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884347962746949e-05,
+      "loss": 0.9882,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878603353545854e-05,
+      "loss": 0.8836,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872857237342067e-05,
+      "loss": 0.9351,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3867109616368208e-05,
+      "loss": 0.9985,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861360492857464e-05,
+      "loss": 1.0064,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855609869043618e-05,
+      "loss": 0.978,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849857747161036e-05,
+      "loss": 0.9726,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844104129444655e-05,
+      "loss": 0.9636,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838349018130007e-05,
+      "loss": 0.8982,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832592415453193e-05,
+      "loss": 0.9932,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 1.0693,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3821074744960394e-05,
+      "loss": 1.0121,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3815313681619515e-05,
+      "loss": 0.9264,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3809551135866683e-05,
+      "loss": 0.9304,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3803787109940891e-05,
+      "loss": 0.9539,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3798021606081713e-05,
+      "loss": 0.9508,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3792254626529286e-05,
+      "loss": 0.8981,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786486173524331e-05,
+      "loss": 0.985,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378071624930814e-05,
+      "loss": 0.9909,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377494485612257e-05,
+      "loss": 1.0425,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769171996210053e-05,
+      "loss": 0.9675,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763397671813597e-05,
+      "loss": 0.9698,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3757621885176768e-05,
+      "loss": 1.0149,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375184463854371e-05,
+      "loss": 1.0549,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746065934159123e-05,
+      "loss": 0.9598,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740285774268282e-05,
+      "loss": 1.0089,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734504161117028e-05,
+      "loss": 0.9074,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728721096951762e-05,
+      "loss": 0.961,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722936584019453e-05,
+      "loss": 0.9584,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717150624567627e-05,
+      "loss": 1.0117,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371136322084438e-05,
+      "loss": 1.0013,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705574375098365e-05,
+      "loss": 0.9798,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699784089578791e-05,
+      "loss": 1.0127,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693992366535433e-05,
+      "loss": 0.9928,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688199208218623e-05,
+      "loss": 1.0294,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682404616879246e-05,
+      "loss": 0.9657,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676608594768754e-05,
+      "loss": 0.9688,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670811144139144e-05,
+      "loss": 0.9648,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665012267242974e-05,
+      "loss": 0.9539,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659211966333357e-05,
+      "loss": 0.9254,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 1.0049,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364760710148898e-05,
+      "loss": 0.9708,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641802542063208e-05,
+      "loss": 0.9791,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635996567641952e-05,
+      "loss": 1.015,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3630189180481083e-05,
+      "loss": 1.008,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624380382837017e-05,
+      "loss": 0.988,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618570176966723e-05,
+      "loss": 0.8879,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361275856512771e-05,
+      "loss": 1.0127,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606945549578039e-05,
+      "loss": 0.882,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360113113257631e-05,
+      "loss": 1.0176,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595315316381676e-05,
+      "loss": 0.9603,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589498103253828e-05,
+      "loss": 0.9122,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3583679495453e-05,
+      "loss": 0.9628,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3577859495239975e-05,
+      "loss": 0.9998,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572038104876066e-05,
+      "loss": 1.0295,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566215326623131e-05,
+      "loss": 1.0047,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356039116274357e-05,
+      "loss": 0.9278,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554565615500313e-05,
+      "loss": 1.0766,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548738687156839e-05,
+      "loss": 0.9348,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542910379977158e-05,
+      "loss": 0.9799,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537080696225815e-05,
+      "loss": 0.9408,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353124963816789e-05,
+      "loss": 0.8979,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525417208068996e-05,
+      "loss": 0.9762,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3519583408195283e-05,
+      "loss": 1.0048,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3513748240813429e-05,
+      "loss": 0.9787,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3507911708190646e-05,
+      "loss": 0.9944,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3502073812594677e-05,
+      "loss": 1.0545,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.349623455629379e-05,
+      "loss": 0.9838,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3490393941556787e-05,
+      "loss": 0.9989,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484551970652997e-05,
+      "loss": 0.891,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478708645852272e-05,
+      "loss": 0.9861,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472863969424998e-05,
+      "loss": 1.0067,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467017943642074e-05,
+      "loss": 0.9133,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346117057077493e-05,
+      "loss": 0.9541,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455321853095527e-05,
+      "loss": 0.9713,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449471792876333e-05,
+      "loss": 0.9835,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3443620392390352e-05,
+      "loss": 1.0167,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437767653911097e-05,
+      "loss": 1.0143,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431913579712612e-05,
+      "loss": 1.0236,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426058172069451e-05,
+      "loss": 0.9699,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.9606,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341434336554992e-05,
+      "loss": 0.9733,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340848397122525e-05,
+      "loss": 0.9573,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402623252559308e-05,
+      "loss": 0.9796,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396761211829229e-05,
+      "loss": 0.9574,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390897851312667e-05,
+      "loss": 0.9925,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338503317328779e-05,
+      "loss": 1.0066,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379167180033272e-05,
+      "loss": 1.0232,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373299873828303e-05,
+      "loss": 0.8408,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336743125695258e-05,
+      "loss": 1.0062,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336156133168631e-05,
+      "loss": 0.9813,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355690100310213e-05,
+      "loss": 0.9908,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349817565105507e-05,
+      "loss": 0.9401,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334394372835393e-05,
+      "loss": 0.9947,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333806859233771e-05,
+      "loss": 0.9776,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332192159339595e-05,
+      "loss": 0.9583,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326314431642821e-05,
+      "loss": 0.9248,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320435411531145e-05,
+      "loss": 0.974,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314555101288808e-05,
+      "loss": 0.9732,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308673503200566e-05,
+      "loss": 0.9461,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.9113,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296906452627874e-05,
+      "loss": 0.9835,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291021004715423e-05,
+      "loss": 1.0338,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285134278101065e-05,
+      "loss": 0.9919,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279246275072046e-05,
+      "loss": 1.0008,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273356997916106e-05,
+      "loss": 0.9859,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267466448921475e-05,
+      "loss": 0.9734,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326157463037689e-05,
+      "loss": 0.8946,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3255681544571568e-05,
+      "loss": 0.9019,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249787193795227e-05,
+      "loss": 0.9744,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243891580338074e-05,
+      "loss": 0.9443,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237994706490802e-05,
+      "loss": 0.9468,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232096574544602e-05,
+      "loss": 1.0303,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3226197186791147e-05,
+      "loss": 0.9327,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220296545522601e-05,
+      "loss": 1.0063,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3214394653031616e-05,
+      "loss": 0.9944,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.320849151161133e-05,
+      "loss": 0.9701,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.320258712355536e-05,
+      "loss": 0.9964,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3196681491157816e-05,
+      "loss": 1.0046,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3190774616713286e-05,
+      "loss": 1.0028,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3184866502516846e-05,
+      "loss": 0.9491,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178957150864047e-05,
+      "loss": 0.9718,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173046564050923e-05,
+      "loss": 0.9635,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167134744373993e-05,
+      "loss": 0.9844,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161221694130247e-05,
+      "loss": 1.0545,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155307415617156e-05,
+      "loss": 0.9158,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149391911132674e-05,
+      "loss": 0.9739,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3143475182975222e-05,
+      "loss": 1.0018,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3137557233443707e-05,
+      "loss": 0.9048,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131638064837496e-05,
+      "loss": 0.9808,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125717679456447e-05,
+      "loss": 0.935,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3119796079600875e-05,
+      "loss": 0.9873,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113873267571577e-05,
+      "loss": 0.9698,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310794924566982e-05,
+      "loss": 0.9206,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102024016197337e-05,
+      "loss": 0.9813,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309609758145633e-05,
+      "loss": 0.9446,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.9967,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084241105379909e-05,
+      "loss": 0.9788,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078311068651241e-05,
+      "loss": 0.9476,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307237983586754e-05,
+      "loss": 0.9365,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066447409333345e-05,
+      "loss": 1.0136,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060513791353655e-05,
+      "loss": 0.949,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054578984233935e-05,
+      "loss": 0.9114,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048642990280109e-05,
+      "loss": 0.9249,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042705811798565e-05,
+      "loss": 0.9748,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036767451096148e-05,
+      "loss": 1.0026,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030827910480167e-05,
+      "loss": 1.0244,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3024887192258384e-05,
+      "loss": 0.9452,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018945298739022e-05,
+      "loss": 1.0108,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301300223223076e-05,
+      "loss": 0.9506,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300705799504273e-05,
+      "loss": 1.017,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300111258948453e-05,
+      "loss": 0.9956,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2995166017866194e-05,
+      "loss": 0.9368,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2989218282498222e-05,
+      "loss": 0.9359,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2983269385691562e-05,
+      "loss": 0.996,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977319329757616e-05,
+      "loss": 0.9356,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971368117008232e-05,
+      "loss": 0.942,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296541574975571e-05,
+      "loss": 0.9481,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29594622303128e-05,
+      "loss": 0.982,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953507560992696e-05,
+      "loss": 0.9231,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2947551744109044e-05,
+      "loss": 1.0136,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941594781975928e-05,
+      "loss": 0.98,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935636676907886e-05,
+      "loss": 1.0172,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292967743121989e-05,
+      "loss": 0.9649,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923717047227368e-05,
+      "loss": 1.0085,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917755527246179e-05,
+      "loss": 1.0087,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911792873592628e-05,
+      "loss": 0.9967,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2905829088583462e-05,
+      "loss": 0.9902,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2899864174535863e-05,
+      "loss": 0.9823,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2893898133767454e-05,
+      "loss": 1.0307,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28879309685963e-05,
+      "loss": 0.9727,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2881962681340894e-05,
+      "loss": 0.941,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2875993274320173e-05,
+      "loss": 0.9903,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870022749853508e-05,
+      "loss": 0.9223,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864051110260697e-05,
+      "loss": 0.9236,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2858078357861979e-05,
+      "loss": 0.9129,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852104494978024e-05,
+      "loss": 0.9698,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846129523929929e-05,
+      "loss": 0.9393,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284015344703923e-05,
+      "loss": 0.9451,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283417626662788e-05,
+      "loss": 0.9562,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828197985018276e-05,
+      "loss": 1.0146,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822218604533232e-05,
+      "loss": 0.9477,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281623812749599e-05,
+      "loss": 0.9349,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810256556230226e-05,
+      "loss": 0.9939,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804273893060028e-05,
+      "loss": 0.9593,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798290140309924e-05,
+      "loss": 0.9442,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792305300304848e-05,
+      "loss": 0.9422,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278631937537017e-05,
+      "loss": 0.8867,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780332367831678e-05,
+      "loss": 0.9836,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774344280015578e-05,
+      "loss": 1.0044,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768355114248493e-05,
+      "loss": 0.9748,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276236487285748e-05,
+      "loss": 0.951,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756373558169992e-05,
+      "loss": 0.9495,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750381172513918e-05,
+      "loss": 0.9248,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274438771821755e-05,
+      "loss": 0.9343,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738393197609602e-05,
+      "loss": 0.9621,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732397613019203e-05,
+      "loss": 0.9233,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272640096677589e-05,
+      "loss": 0.986,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720403261209615e-05,
+      "loss": 0.9912,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714404498650743e-05,
+      "loss": 1.031,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708404681430054e-05,
+      "loss": 0.9843,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702403811878727e-05,
+      "loss": 0.9682,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696401892328356e-05,
+      "loss": 0.9538,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690398925110943e-05,
+      "loss": 0.9942,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684394912558898e-05,
+      "loss": 0.871,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678389857005033e-05,
+      "loss": 0.9926,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267238376078257e-05,
+      "loss": 0.9473,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666376626225131e-05,
+      "loss": 0.9592,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660368455666752e-05,
+      "loss": 1.0431,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654359251441854e-05,
+      "loss": 0.9971,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648349015885272e-05,
+      "loss": 0.9689,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642337751332238e-05,
+      "loss": 1.0456,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636325460118388e-05,
+      "loss": 0.928,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630312144579748e-05,
+      "loss": 0.9364,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262429780705275e-05,
+      "loss": 0.8903,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618282449874221e-05,
+      "loss": 0.9658,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612266075381385e-05,
+      "loss": 0.9027,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606248685911856e-05,
+      "loss": 0.9272,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600230283803648e-05,
+      "loss": 0.8978,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2594210871395164e-05,
+      "loss": 0.9315,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 1.0047,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2582169025032966e-05,
+      "loss": 1.0091,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2576146595758019e-05,
+      "loss": 1.0023,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2570123165540338e-05,
+      "loss": 0.9926,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2564098736720286e-05,
+      "loss": 0.965,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558073311638604e-05,
+      "loss": 1.0011,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552046892636427e-05,
+      "loss": 0.9892,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546019482055276e-05,
+      "loss": 0.9955,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539991082237062e-05,
+      "loss": 0.9669,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533961695524071e-05,
+      "loss": 0.9828,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527931324258975e-05,
+      "loss": 0.9927,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521899970784834e-05,
+      "loss": 0.9744,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515867637445088e-05,
+      "loss": 0.9554,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2509834326583548e-05,
+      "loss": 0.9882,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2503800040544417e-05,
+      "loss": 0.9242,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249776478167227e-05,
+      "loss": 0.9646,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491728552312066e-05,
+      "loss": 0.9673,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2485691354809134e-05,
+      "loss": 0.9949,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2479653191509182e-05,
+      "loss": 0.9893,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473614064758297e-05,
+      "loss": 0.9958,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467573976902936e-05,
+      "loss": 0.9043,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461532930289932e-05,
+      "loss": 0.9278,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2455490927266485e-05,
+      "loss": 0.9427,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449447970180172e-05,
+      "loss": 0.966,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443404061378941e-05,
+      "loss": 0.9439,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437359203211109e-05,
+      "loss": 0.982,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431313398025356e-05,
+      "loss": 0.9245,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242526664817074e-05,
+      "loss": 0.934,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419218955996677e-05,
+      "loss": 0.9628,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2413170323852959e-05,
+      "loss": 0.9387,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407120754089733e-05,
+      "loss": 0.9416,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401070249057514e-05,
+      "loss": 0.9257,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239501881110718e-05,
+      "loss": 1.0137,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388966442589978e-05,
+      "loss": 0.9906,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382913145857507e-05,
+      "loss": 1.0312,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376858923261732e-05,
+      "loss": 0.9139,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370803777154976e-05,
+      "loss": 1.007,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364747709889923e-05,
+      "loss": 0.9809,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358690723819611e-05,
+      "loss": 0.978,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352632821297437e-05,
+      "loss": 0.884,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346574004677154e-05,
+      "loss": 1.0105,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340514276312874e-05,
+      "loss": 0.9283,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2334453638559057e-05,
+      "loss": 0.9893,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2328392093770517e-05,
+      "loss": 0.9504,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322329644302426e-05,
+      "loss": 0.9643,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316266292510305e-05,
+      "loss": 0.9857,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310202040750022e-05,
+      "loss": 0.9645,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304136891377795e-05,
+      "loss": 0.9226,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298070846750197e-05,
+      "loss": 0.9551,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292003909224144e-05,
+      "loss": 0.8664,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285936081156897e-05,
+      "loss": 0.9779,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.227986736490607e-05,
+      "loss": 0.9626,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2273797762829615e-05,
+      "loss": 0.9768,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2267727277285832e-05,
+      "loss": 0.9592,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2261655910633365e-05,
+      "loss": 1.0107,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2255583665231196e-05,
+      "loss": 1.0466,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2249510543438652e-05,
+      "loss": 0.9336,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22434365476154e-05,
+      "loss": 0.9775,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237361680121445e-05,
+      "loss": 0.9374,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231285943317132e-05,
+      "loss": 0.9857,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.9475,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219131871220504e-05,
+      "loss": 0.935,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213053540650566e-05,
+      "loss": 0.9418,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206974350215016e-05,
+      "loss": 0.9781,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200894302275878e-05,
+      "loss": 0.9212,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194813399195518e-05,
+      "loss": 0.9853,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2188731643336615e-05,
+      "loss": 0.9822,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182649037062203e-05,
+      "loss": 0.8706,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176565582735624e-05,
+      "loss": 0.8423,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170481282720563e-05,
+      "loss": 0.9957,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164396139381029e-05,
+      "loss": 0.899,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215831015508136e-05,
+      "loss": 0.9773,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152223332186222e-05,
+      "loss": 0.8555,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146135673060599e-05,
+      "loss": 0.9758,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140047180069814e-05,
+      "loss": 0.9486,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133957855579501e-05,
+      "loss": 0.9688,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127867701955622e-05,
+      "loss": 0.9723,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121776721564465e-05,
+      "loss": 0.9501,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211568491677263e-05,
+      "loss": 0.8897,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109592289947045e-05,
+      "loss": 0.9803,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210349884345496e-05,
+      "loss": 0.9475,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209740457966393e-05,
+      "loss": 0.9399,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091309500941841e-05,
+      "loss": 1.0197,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085213609656891e-05,
+      "loss": 0.9109,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 0.9948,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073019398872778e-05,
+      "loss": 0.9782,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066921084111582e-05,
+      "loss": 0.9504,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2060821966263465e-05,
+      "loss": 1.0476,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054722047698192e-05,
+      "loss": 0.9977,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048621330785842e-05,
+      "loss": 0.9537,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042519817896805e-05,
+      "loss": 0.9683,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036417511401775e-05,
+      "loss": 0.9522,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030314413671763e-05,
+      "loss": 1.0153,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202421052707808e-05,
+      "loss": 1.046,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018105853992346e-05,
+      "loss": 0.9656,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012000396786485e-05,
+      "loss": 0.9463,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200589415783273e-05,
+      "loss": 0.9158,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999787139503614e-05,
+      "loss": 0.9732,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1993679344171973e-05,
+      "loss": 0.9271,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987570774210948e-05,
+      "loss": 0.8973,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981461431993978e-05,
+      "loss": 1.0039,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975351319894801e-05,
+      "loss": 0.9416,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969240440287457e-05,
+      "loss": 1.0305,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1963128795546286e-05,
+      "loss": 0.9234,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1957016388045917e-05,
+      "loss": 1.003,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.9199,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1944789294267614e-05,
+      "loss": 0.9635,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1938674612740424e-05,
+      "loss": 0.9518,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1932559177955533e-05,
+      "loss": 0.9538,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926442992289047e-05,
+      "loss": 0.998,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920326058117364e-05,
+      "loss": 0.9709,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914208377817171e-05,
+      "loss": 0.9108,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190808995376545e-05,
+      "loss": 0.9746,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190197078833947e-05,
+      "loss": 0.9724,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1895850883916786e-05,
+      "loss": 0.9976,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1889730242875243e-05,
+      "loss": 0.8918,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188360886759297e-05,
+      "loss": 0.9407,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877486760448382e-05,
+      "loss": 0.972,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1871363923820182e-05,
+      "loss": 1.0356,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865240360087349e-05,
+      "loss": 1.0063,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859116071629148e-05,
+      "loss": 0.9063,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852991060825128e-05,
+      "loss": 1.0137,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846865330055115e-05,
+      "loss": 0.9668,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1840738881699218e-05,
+      "loss": 0.9453,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834611718137825e-05,
+      "loss": 0.9957,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1828483841751597e-05,
+      "loss": 0.9769,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1822355254921478e-05,
+      "loss": 0.9533,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181622596002868e-05,
+      "loss": 0.9146,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18100959594547e-05,
+      "loss": 0.9486,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803965255581304e-05,
+      "loss": 0.9453,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797833850790527e-05,
+      "loss": 0.9152,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179170174746469e-05,
+      "loss": 0.9054,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1785568947986368e-05,
+      "loss": 0.9488,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177943545473842e-05,
+      "loss": 0.9289,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773301270103968e-05,
+      "loss": 0.9725,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767166396466404e-05,
+      "loss": 0.9638,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761030836209384e-05,
+      "loss": 1.0329,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754894591716841e-05,
+      "loss": 1.0287,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1748757665372965e-05,
+      "loss": 0.9102,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1742620059562211e-05,
+      "loss": 0.9313,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.9883,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730342819079229e-05,
+      "loss": 0.9681,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1724203189177228e-05,
+      "loss": 0.8987,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718062889348816e-05,
+      "loss": 0.9167,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711921921979754e-05,
+      "loss": 0.8629,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705780289456069e-05,
+      "loss": 0.8891,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699637994164053e-05,
+      "loss": 0.955,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1693495038490247e-05,
+      "loss": 0.9097,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1687351424821448e-05,
+      "loss": 0.9729,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168120715554472e-05,
+      "loss": 0.9711,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675062233047365e-05,
+      "loss": 0.992,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166891665971695e-05,
+      "loss": 0.9678,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662770437941293e-05,
+      "loss": 1.0187,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165662357010846e-05,
+      "loss": 0.9667,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650476058606776e-05,
+      "loss": 0.9796,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1644327905824808e-05,
+      "loss": 0.9258,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1638179114151378e-05,
+      "loss": 0.9808,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.163202968597555e-05,
+      "loss": 0.9649,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.162587962368664e-05,
+      "loss": 0.9885,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.161972892967421e-05,
+      "loss": 0.8794,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1613577606328068e-05,
+      "loss": 0.9776,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1607425656038263e-05,
+      "loss": 0.9492,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601273081195087e-05,
+      "loss": 1.0081,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595119884189085e-05,
+      "loss": 0.9845,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588966067411033e-05,
+      "loss": 0.8901,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582811633251949e-05,
+      "loss": 1.0174,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1576656584103097e-05,
+      "loss": 0.9497,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570500922355971e-05,
+      "loss": 0.9897,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 0.9332,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558187770634089e-05,
+      "loss": 1.0062,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1552030285443516e-05,
+      "loss": 0.9866,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545872197223039e-05,
+      "loss": 0.94,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1539713508365336e-05,
+      "loss": 0.9926,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153355422126332e-05,
+      "loss": 0.9729,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152739433831014e-05,
+      "loss": 0.9648,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521233861899168e-05,
+      "loss": 1.0138,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515072794424013e-05,
+      "loss": 0.9267,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508911138278516e-05,
+      "loss": 0.9348,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502748895856737e-05,
+      "loss": 0.9901,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496586069552973e-05,
+      "loss": 0.9802,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490422661761744e-05,
+      "loss": 0.9393,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14842586748778e-05,
+      "loss": 0.9569,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478094111296109e-05,
+      "loss": 0.9757,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471928973411865e-05,
+      "loss": 0.9195,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146576326362049e-05,
+      "loss": 0.9773,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459596984317622e-05,
+      "loss": 1.0062,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1453430137899129e-05,
+      "loss": 0.9061,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447262726761088e-05,
+      "loss": 0.9485,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441094753299802e-05,
+      "loss": 0.97,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434926219911792e-05,
+      "loss": 1.0136,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428757128993801e-05,
+      "loss": 0.9849,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422587482942778e-05,
+      "loss": 0.9609,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416417284155892e-05,
+      "loss": 0.9834,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410246535030533e-05,
+      "loss": 0.9766,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404075237964296e-05,
+      "loss": 0.9612,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397903395354996e-05,
+      "loss": 0.9723,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391731009600655e-05,
+      "loss": 0.985,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138555808309951e-05,
+      "loss": 0.9409,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379384618250002e-05,
+      "loss": 0.968,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373210617450793e-05,
+      "loss": 0.9185,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367036083100735e-05,
+      "loss": 0.8918,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1360861017598905e-05,
+      "loss": 0.9769,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354685423344577e-05,
+      "loss": 0.9657,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348509302737232e-05,
+      "loss": 0.9944,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342332658176556e-05,
+      "loss": 0.8671,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1336155492062439e-05,
+      "loss": 0.9755,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329977806794976e-05,
+      "loss": 0.9258,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1323799604774457e-05,
+      "loss": 0.9493,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1317620888401379e-05,
+      "loss": 0.9415,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1311441660076433e-05,
+      "loss": 0.9529,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.130526192220052e-05,
+      "loss": 0.9493,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1299081677174721e-05,
+      "loss": 1.026,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1292900927400334e-05,
+      "loss": 0.973,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1286719675278838e-05,
+      "loss": 0.9563,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280537923211919e-05,
+      "loss": 0.9375,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274355673601446e-05,
+      "loss": 0.9835,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268172928849486e-05,
+      "loss": 0.9934,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12619896913583e-05,
+      "loss": 0.9892,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255805963530339e-05,
+      "loss": 0.9156,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249621747768247e-05,
+      "loss": 0.9854,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243437046474854e-05,
+      "loss": 0.9921,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237251862053178e-05,
+      "loss": 1.0067,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123106619690643e-05,
+      "loss": 0.9445,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224880053438006e-05,
+      "loss": 0.9273,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218693434051475e-05,
+      "loss": 0.9728,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212506341150615e-05,
+      "loss": 0.97,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120631877713937e-05,
+      "loss": 1.0335,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200130744421877e-05,
+      "loss": 0.9857,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193942245402443e-05,
+      "loss": 0.9713,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187753282485569e-05,
+      "loss": 1.0083,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118156385807593e-05,
+      "loss": 0.9569,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175373974578378e-05,
+      "loss": 0.9672,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1169183634397948e-05,
+      "loss": 0.948,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162992839939852e-05,
+      "loss": 0.9645,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156801593609479e-05,
+      "loss": 0.9505,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1150609897812387e-05,
+      "loss": 0.9877,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114441775495432e-05,
+      "loss": 0.9,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138225167441186e-05,
+      "loss": 0.9676,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113203213767907e-05,
+      "loss": 0.954,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1125838668074225e-05,
+      "loss": 0.8655,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.8767,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113450418962233e-05,
+      "loss": 0.981,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107255644268448e-05,
+      "loss": 0.9416,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110106043935866e-05,
+      "loss": 0.9376,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094864806639971e-05,
+      "loss": 0.9609,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088668748519646e-05,
+      "loss": 0.8962,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082472267405123e-05,
+      "loss": 0.9373,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107627536570399e-05,
+      "loss": 0.9484,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070078045824014e-05,
+      "loss": 0.9917,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1063880310173114e-05,
+      "loss": 0.9885,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105768216115938e-05,
+      "loss": 0.9877,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1051483601191051e-05,
+      "loss": 0.9173,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.9635,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039085258024396e-05,
+      "loss": 0.9909,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032885479643351e-05,
+      "loss": 1.0489,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026685299942286e-05,
+      "loss": 0.9577,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020484721330227e-05,
+      "loss": 0.9635,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1014283746216366e-05,
+      "loss": 0.9773,
+      "step": 2519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008082377010045e-05,
+      "loss": 0.9912,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1001880616120764e-05,
+      "loss": 0.9651,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0995678465958168e-05,
+      "loss": 0.8981,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0989475928932057e-05,
+      "loss": 1.0205,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.098327300745238e-05,
+      "loss": 0.9425,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0977069703929242e-05,
+      "loss": 0.9809,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0970866020772884e-05,
+      "loss": 1.0083,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964661960393703e-05,
+      "loss": 0.9921,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958457525202241e-05,
+      "loss": 0.8901,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0952252717609188e-05,
+      "loss": 0.9572,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946047540025373e-05,
+      "loss": 0.953,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939841994861773e-05,
+      "loss": 0.9152,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933636084529507e-05,
+      "loss": 0.9791,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0927429811439838e-05,
+      "loss": 0.9752,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921223178004163e-05,
+      "loss": 0.8922,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915016186634027e-05,
+      "loss": 0.9514,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908808839741108e-05,
+      "loss": 0.9888,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902601139737225e-05,
+      "loss": 0.9965,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896393089034336e-05,
+      "loss": 0.9529,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890184690044534e-05,
+      "loss": 0.9515,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0883975945180043e-05,
+      "loss": 0.9298,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877766856853227e-05,
+      "loss": 0.9383,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871557427476585e-05,
+      "loss": 0.9427,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865347659462737e-05,
+      "loss": 0.9113,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859137555224448e-05,
+      "loss": 0.9403,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852927117174607e-05,
+      "loss": 0.9341,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846716347726233e-05,
+      "loss": 0.9953,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840505249292477e-05,
+      "loss": 0.9929,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834293824286612e-05,
+      "loss": 0.9181,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828082075122044e-05,
+      "loss": 0.9678,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821870004212305e-05,
+      "loss": 0.8958,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081565761397104e-05,
+      "loss": 0.9709,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809444906812034e-05,
+      "loss": 0.9475,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0803231885149191e-05,
+      "loss": 0.8794,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797018551396527e-05,
+      "loss": 1.0206,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790804907968194e-05,
+      "loss": 1.0112,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0784590957278452e-05,
+      "loss": 0.8516,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0778376701741688e-05,
+      "loss": 0.9622,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772162143772407e-05,
+      "loss": 0.9419,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765947285785228e-05,
+      "loss": 0.9318,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759732130194887e-05,
+      "loss": 0.878,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753516679416237e-05,
+      "loss": 0.9994,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.8838,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741084901953995e-05,
+      "loss": 0.9342,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734868580100677e-05,
+      "loss": 0.958,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728651972719603e-05,
+      "loss": 0.9926,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722435082226186e-05,
+      "loss": 0.88,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0716217911035952e-05,
+      "loss": 0.9554,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0710000461564535e-05,
+      "loss": 0.9801,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703782736227688e-05,
+      "loss": 0.9807,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697564737441254e-05,
+      "loss": 0.9536,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691346467621192e-05,
+      "loss": 0.9444,
+      "step": 2571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685127929183567e-05,
+      "loss": 0.9724,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0678909124544548e-05,
+      "loss": 0.9744,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0672690056120398e-05,
+      "loss": 0.9677,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0666470726327496e-05,
+      "loss": 1.0367,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0660251137582316e-05,
+      "loss": 0.9744,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0654031292301432e-05,
+      "loss": 0.9382,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0647811192901518e-05,
+      "loss": 0.9208,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641590841799352e-05,
+      "loss": 0.982,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635370241411803e-05,
+      "loss": 0.9216,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629149394155837e-05,
+      "loss": 0.9332,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622928302448523e-05,
+      "loss": 0.9699,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616706968707015e-05,
+      "loss": 0.9216,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610485395348571e-05,
+      "loss": 1.0084,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604263584790534e-05,
+      "loss": 0.9715,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598041539450344e-05,
+      "loss": 0.9748,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591819261745528e-05,
+      "loss": 0.9614,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585596754093712e-05,
+      "loss": 0.9671,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579374018912602e-05,
+      "loss": 0.9311,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573151058619994e-05,
+      "loss": 0.9635,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566927875633776e-05,
+      "loss": 1.0167,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560704472371919e-05,
+      "loss": 0.981,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055448085125248e-05,
+      "loss": 0.969,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548257014693602e-05,
+      "loss": 0.9678,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542032965113508e-05,
+      "loss": 0.9918,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053580870493051e-05,
+      "loss": 0.9412,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529584236562995e-05,
+      "loss": 0.9531,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0523359562429441e-05,
+      "loss": 0.9345,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051713468494839e-05,
+      "loss": 0.9178,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510909606538478e-05,
+      "loss": 0.9242,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504684329618408e-05,
+      "loss": 0.9614,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498458856606972e-05,
+      "loss": 0.9253,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049223318992303e-05,
+      "loss": 0.9807,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486007331985516e-05,
+      "loss": 0.8688,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047978128521344e-05,
+      "loss": 0.968,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473555052025893e-05,
+      "loss": 0.9313,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467328634842024e-05,
+      "loss": 0.8855,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461102036081069e-05,
+      "loss": 1.0004,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045487525816232e-05,
+      "loss": 1.0378,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044864830350515e-05,
+      "loss": 0.9665,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442421174528997e-05,
+      "loss": 0.9799,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436193873653362e-05,
+      "loss": 0.9455,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429966403297818e-05,
+      "loss": 0.8994,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423738765882006e-05,
+      "loss": 0.9915,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417510963825625e-05,
+      "loss": 0.9708,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411282999548442e-05,
+      "loss": 0.9948,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405054875470287e-05,
+      "loss": 0.958,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039882659401105e-05,
+      "loss": 0.9819,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0392598157590687e-05,
+      "loss": 0.9872,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386369568629209e-05,
+      "loss": 0.9355,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038014082954669e-05,
+      "loss": 0.9899,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037391194276326e-05,
+      "loss": 0.9669,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367682910699104e-05,
+      "loss": 0.9108,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0361453735774467e-05,
+      "loss": 0.9557,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0355224420409653e-05,
+      "loss": 0.9559,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0348994967025012e-05,
+      "loss": 0.9819,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0342765378040953e-05,
+      "loss": 0.9594,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0336535655877942e-05,
+      "loss": 0.9433,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0330305802956484e-05,
+      "loss": 0.9889,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0324075821697146e-05,
+      "loss": 0.9544,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0317845714520543e-05,
+      "loss": 0.9551,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311615483847333e-05,
+      "loss": 0.9715,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305385132098229e-05,
+      "loss": 0.9989,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299154661693987e-05,
+      "loss": 0.9528,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029292407505541e-05,
+      "loss": 0.96,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286693374603347e-05,
+      "loss": 0.9583,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028046256275869e-05,
+      "loss": 0.9895,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274231641942378e-05,
+      "loss": 1.0103,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268000614575386e-05,
+      "loss": 0.9769,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261769483078734e-05,
+      "loss": 0.9294,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255538249873482e-05,
+      "loss": 0.9537,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249306917380731e-05,
+      "loss": 1.0107,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243075488021617e-05,
+      "loss": 0.8779,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236843964217317e-05,
+      "loss": 1.0345,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0230612348389048e-05,
+      "loss": 0.9886,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224380642958052e-05,
+      "loss": 0.9812,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0218148850345613e-05,
+      "loss": 0.9777,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211916972973048e-05,
+      "loss": 0.9661,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205685013261712e-05,
+      "loss": 0.9969,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199452973632982e-05,
+      "loss": 0.9525,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019322085650827e-05,
+      "loss": 0.9325,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0186988664309023e-05,
+      "loss": 0.9415,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018075639945671e-05,
+      "loss": 0.8985,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174524064372837e-05,
+      "loss": 0.9242,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168291661478924e-05,
+      "loss": 1.0009,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016205919319653e-05,
+      "loss": 0.9262,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155826661947232e-05,
+      "loss": 0.9377,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149594070152638e-05,
+      "loss": 0.9686,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014336142023437e-05,
+      "loss": 0.9189,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137128714614082e-05,
+      "loss": 1.0064,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130895955713445e-05,
+      "loss": 0.9971,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124663145954152e-05,
+      "loss": 0.9739,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118430287757914e-05,
+      "loss": 0.9919,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011219738354646e-05,
+      "loss": 0.9874,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105964435741544e-05,
+      "loss": 0.9558,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099731446764927e-05,
+      "loss": 0.9949,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093498419038394e-05,
+      "loss": 0.972,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008726535498374e-05,
+      "loss": 1.0188,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081032257022776e-05,
+      "loss": 0.9052,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007479912757733e-05,
+      "loss": 0.8711,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068565969069236e-05,
+      "loss": 0.9147,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062332783920337e-05,
+      "loss": 0.9647,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056099574552498e-05,
+      "loss": 0.9582,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049866343387582e-05,
+      "loss": 0.958,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043633092847468e-05,
+      "loss": 0.9564,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0037399825354036e-05,
+      "loss": 0.9566,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0031166543329179e-05,
+      "loss": 0.88,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0024933249194792e-05,
+      "loss": 0.9822,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0018699945372776e-05,
+      "loss": 0.8976,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0012466634285034e-05,
+      "loss": 0.9798,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0006233318353474e-05,
+      "loss": 0.9431,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 1.0554,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993766681646531e-06,
+      "loss": 0.9559,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987533365714968e-06,
+      "loss": 0.9971,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981300054627227e-06,
+      "loss": 0.9228,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97506675080521e-06,
+      "loss": 0.9576,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968833456670824e-06,
+      "loss": 0.9383,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962600174645968e-06,
+      "loss": 0.9404,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956366907152536e-06,
+      "loss": 1.0382,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950133656612421e-06,
+      "loss": 1.0051,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943900425447505e-06,
+      "loss": 0.9892,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937667216079665e-06,
+      "loss": 1.0205,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931434030930768e-06,
+      "loss": 0.988,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925200872422671e-06,
+      "loss": 0.9742,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918967742977226e-06,
+      "loss": 1.0108,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912734645016262e-06,
+      "loss": 0.9848,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90650158096161e-06,
+      "loss": 0.9626,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900268553235077e-06,
+      "loss": 0.9114,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89403556425846e-06,
+      "loss": 0.9498,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887802616453543e-06,
+      "loss": 0.9248,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88156971224209e-06,
+      "loss": 0.9014,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87533685404585e-06,
+      "loss": 0.93,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869104044286558e-06,
+      "loss": 0.9338,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862871285385921e-06,
+      "loss": 0.9219,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856638579765633e-06,
+      "loss": 0.9477,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850405929847367e-06,
+      "loss": 1.0025,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844173338052771e-06,
+      "loss": 0.9977,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837940806803474e-06,
+      "loss": 1.0072,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83170833852108e-06,
+      "loss": 1.0017,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825475935627165e-06,
+      "loss": 0.9603,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819243600543292e-06,
+      "loss": 0.9288,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81301133569098e-06,
+      "loss": 0.9753,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806779143491733e-06,
+      "loss": 0.9702,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800547026367022e-06,
+      "loss": 0.9114,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794314986738291e-06,
+      "loss": 1.0044,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788083027026953e-06,
+      "loss": 0.9988,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78185114965439e-06,
+      "loss": 0.9494,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775619357041952e-06,
+      "loss": 0.9377,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769387651610955e-06,
+      "loss": 0.9409,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763156035782686e-06,
+      "loss": 0.9663,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756924511978388e-06,
+      "loss": 0.9813,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750693082619274e-06,
+      "loss": 0.9007,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744461750126523e-06,
+      "loss": 0.947,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738230516921272e-06,
+      "loss": 0.9681,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73199938542462e-06,
+      "loss": 0.9803,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725768358057625e-06,
+      "loss": 0.9768,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719537437241311e-06,
+      "loss": 1.0105,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.713306625396658e-06,
+      "loss": 1.0014,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.707075924944595e-06,
+      "loss": 0.9953,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.700845338306018e-06,
+      "loss": 0.986,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.694614867901776e-06,
+      "loss": 0.9227,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.68838451615267e-06,
+      "loss": 0.9078,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.68215428547946e-06,
+      "loss": 0.9374,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675924178302857e-06,
+      "loss": 0.941,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66969419704352e-06,
+      "loss": 0.9918,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663464344122064e-06,
+      "loss": 1.0383,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65723462195905e-06,
+      "loss": 0.9314,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651005032974994e-06,
+      "loss": 1.0347,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644775579590352e-06,
+      "loss": 0.9679,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638546264225538e-06,
+      "loss": 0.9227,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632317089300902e-06,
+      "loss": 0.9812,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626088057236745e-06,
+      "loss": 0.9643,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619859170453314e-06,
+      "loss": 0.992,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613630431370794e-06,
+      "loss": 0.9593,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607401842409318e-06,
+      "loss": 0.9553,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601173405988955e-06,
+      "loss": 0.9637,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594945124529718e-06,
+      "loss": 0.9828,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588717000451563e-06,
+      "loss": 0.8746,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58248903617438e-06,
+      "loss": 0.9598,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576261234117998e-06,
+      "loss": 1.0022,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570033596702185e-06,
+      "loss": 0.9376,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563806126346643e-06,
+      "loss": 0.9723,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557578825471008e-06,
+      "loss": 0.8695,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551351696494854e-06,
+      "loss": 0.9754,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545124741837684e-06,
+      "loss": 0.9002,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538897963918936e-06,
+      "loss": 0.8707,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532671365157979e-06,
+      "loss": 1.0047,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526444947974112e-06,
+      "loss": 0.9937,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520218714786564e-06,
+      "loss": 0.9715,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51399266801449e-06,
+      "loss": 0.9185,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507766810076972e-06,
+      "loss": 0.9302,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501541143393028e-06,
+      "loss": 0.9208,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495315670381592e-06,
+      "loss": 0.9976,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489090393461523e-06,
+      "loss": 0.9537,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48286531505161e-06,
+      "loss": 0.9842,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476640437570562e-06,
+      "loss": 0.9145,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470415763437003e-06,
+      "loss": 0.9559,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46419129506949e-06,
+      "loss": 0.9246,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457967034886493e-06,
+      "loss": 0.9294,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4517429853064e-06,
+      "loss": 0.9455,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44551914874752e-06,
+      "loss": 1.018,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439295527628083e-06,
+      "loss": 0.959,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433072124366224e-06,
+      "loss": 0.9565,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426848941380007e-06,
+      "loss": 0.9471,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4206259810874e-06,
+      "loss": 0.9748,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41440324590629e-06,
+      "loss": 0.9147,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408180738254472e-06,
+      "loss": 0.9431,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401958460549658e-06,
+      "loss": 0.9482,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395736415209468e-06,
+      "loss": 0.9719,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.38951460465143e-06,
+      "loss": 1.0134,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.383293031292985e-06,
+      "loss": 0.9663,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.377071697551479e-06,
+      "loss": 1.0342,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.370850605844165e-06,
+      "loss": 0.9612,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.364629758588199e-06,
+      "loss": 0.9754,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.358409158200648e-06,
+      "loss": 1.0374,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.352188807098482e-06,
+      "loss": 0.9112,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34596870769857e-06,
+      "loss": 0.9725,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339748862417686e-06,
+      "loss": 0.9535,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333529273672505e-06,
+      "loss": 0.923,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327309943879604e-06,
+      "loss": 0.8855,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321090875455457e-06,
+      "loss": 0.9366,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314872070816435e-06,
+      "loss": 1.0491,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308653532378809e-06,
+      "loss": 0.962,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302435262558748e-06,
+      "loss": 1.0342,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296217263772313e-06,
+      "loss": 0.9684,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289999538435463e-06,
+      "loss": 0.9639,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28378208896405e-06,
+      "loss": 0.9595,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277564917773816e-06,
+      "loss": 0.9381,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2713480272804e-06,
+      "loss": 0.9715,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265131419899325e-06,
+      "loss": 0.9783,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258915098046008e-06,
+      "loss": 0.9423,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 0.9889,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246483320583768e-06,
+      "loss": 0.9374,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240267869805118e-06,
+      "loss": 0.9666,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234052714214777e-06,
+      "loss": 0.9406,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227837856227594e-06,
+      "loss": 0.9987,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221623298258315e-06,
+      "loss": 0.9452,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.215409042721553e-06,
+      "loss": 0.92,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209195092031811e-06,
+      "loss": 0.8969,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202981448603477e-06,
+      "loss": 0.9118,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196768114850814e-06,
+      "loss": 1.0202,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190555093187968e-06,
+      "loss": 0.9873,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184342386028963e-06,
+      "loss": 0.9631,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178129995787698e-06,
+      "loss": 0.9244,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17191792487796e-06,
+      "loss": 0.9821,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165706175713391e-06,
+      "loss": 0.9191,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159494750707527e-06,
+      "loss": 0.931,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153283652273768e-06,
+      "loss": 0.9215,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147072882825394e-06,
+      "loss": 0.9979,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140862444775554e-06,
+      "loss": 0.9418,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134652340537265e-06,
+      "loss": 0.9491,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128442572523418e-06,
+      "loss": 1.0358,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122233143146777e-06,
+      "loss": 0.9561,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11602405481996e-06,
+      "loss": 0.9065,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10981530995547e-06,
+      "loss": 0.9493,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103606910965666e-06,
+      "loss": 0.9594,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097398860262777e-06,
+      "loss": 1.0004,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091191160258895e-06,
+      "loss": 0.9478,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084983813365977e-06,
+      "loss": 0.8841,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078776821995839e-06,
+      "loss": 1.0002,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072570188560164e-06,
+      "loss": 0.98,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.066363915470494e-06,
+      "loss": 0.9641,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.060158005138229e-06,
+      "loss": 0.9609,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.05395245997463e-06,
+      "loss": 0.9856,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.047747282390815e-06,
+      "loss": 0.9688,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.04154247479776e-06,
+      "loss": 0.9985,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.0353380396063e-06,
+      "loss": 0.9551,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02913397922712e-06,
+      "loss": 0.9656,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022930296070761e-06,
+      "loss": 0.9439,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016726992547621e-06,
+      "loss": 0.9929,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010524071067947e-06,
+      "loss": 1.0052,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004321534041836e-06,
+      "loss": 0.917,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99811938387924e-06,
+      "loss": 0.9076,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991917622989957e-06,
+      "loss": 0.9502,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985716253783636e-06,
+      "loss": 0.9676,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979515278669776e-06,
+      "loss": 0.9775,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973314700057717e-06,
+      "loss": 0.9768,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.967114520356652e-06,
+      "loss": 0.9557,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960914741975608e-06,
+      "loss": 0.9492,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.9711,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948516398808952e-06,
+      "loss": 0.9204,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942317838840625e-06,
+      "loss": 0.9249,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936119689826888e-06,
+      "loss": 0.9279,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92992195417599e-06,
+      "loss": 0.9563,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923724634296013e-06,
+      "loss": 0.9969,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917527732594882e-06,
+      "loss": 1.0137,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911331251480357e-06,
+      "loss": 1.0169,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905135193360032e-06,
+      "loss": 0.985,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898939560641342e-06,
+      "loss": 0.9257,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892744355731555e-06,
+      "loss": 0.9342,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88654958103777e-06,
+      "loss": 0.9433,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.9418,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.874161331925778e-06,
+      "loss": 0.9765,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867967862320935e-06,
+      "loss": 0.9343,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861774832558819e-06,
+      "loss": 0.9491,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855582245045682e-06,
+      "loss": 0.8859,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.849390102187615e-06,
+      "loss": 1.0042,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843198406390524e-06,
+      "loss": 0.9291,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83700716006015e-06,
+      "loss": 0.9243,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830816365602053e-06,
+      "loss": 0.975,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824626025421625e-06,
+      "loss": 0.8469,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818436141924072e-06,
+      "loss": 0.9718,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812246717514434e-06,
+      "loss": 0.9404,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806057754597559e-06,
+      "loss": 0.9155,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799869255578126e-06,
+      "loss": 0.9985,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793681222860631e-06,
+      "loss": 0.9609,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787493658849387e-06,
+      "loss": 0.928,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781306565948528e-06,
+      "loss": 0.8598,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775119946562e-06,
+      "loss": 0.9668,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768933803093573e-06,
+      "loss": 0.974,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.762748137946827e-06,
+      "loss": 0.9727,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.756562953525151e-06,
+      "loss": 0.9339,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750378252231756e-06,
+      "loss": 1.0073,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.744194036469665e-06,
+      "loss": 0.9573,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.738010308641705e-06,
+      "loss": 0.9151,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.731827071150519e-06,
+      "loss": 0.9577,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.72564432639856e-06,
+      "loss": 1.0063,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.719462076788085e-06,
+      "loss": 0.9646,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.713280324721163e-06,
+      "loss": 0.9612,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70709907259967e-06,
+      "loss": 0.9321,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.700918322825282e-06,
+      "loss": 0.8749,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694738077799487e-06,
+      "loss": 0.9544,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68855833992357e-06,
+      "loss": 0.9481,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682379111598626e-06,
+      "loss": 0.9778,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676200395225546e-06,
+      "loss": 0.9194,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670022193205027e-06,
+      "loss": 0.9684,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663844507937563e-06,
+      "loss": 1.0267,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657667341823449e-06,
+      "loss": 0.9582,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651490697262773e-06,
+      "loss": 0.98,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645314576655428e-06,
+      "loss": 0.9532,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6391389824011e-06,
+      "loss": 0.8572,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632963916899268e-06,
+      "loss": 0.8993,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626789382549212e-06,
+      "loss": 0.9265,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62061538175e-06,
+      "loss": 0.9878,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614441916900493e-06,
+      "loss": 0.9774,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60826899039935e-06,
+      "loss": 0.9989,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602096604645009e-06,
+      "loss": 0.9547,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59592476203571e-06,
+      "loss": 1.0201,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589753464969473e-06,
+      "loss": 0.9352,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583582715844113e-06,
+      "loss": 0.9317,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577412517057227e-06,
+      "loss": 0.9511,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571242871006202e-06,
+      "loss": 0.9692,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56507378008821e-06,
+      "loss": 1.0112,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558905246700202e-06,
+      "loss": 0.9641,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552737273238919e-06,
+      "loss": 0.9745,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546569862100876e-06,
+      "loss": 0.9629,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540403015682382e-06,
+      "loss": 0.9265,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534236736379515e-06,
+      "loss": 1.0104,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528071026588137e-06,
+      "loss": 0.9626,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521905888703894e-06,
+      "loss": 0.948,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515741325122202e-06,
+      "loss": 1.0164,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509577338238255e-06,
+      "loss": 0.8656,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503413930447027e-06,
+      "loss": 0.9146,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497251104143263e-06,
+      "loss": 0.9677,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491088861721486e-06,
+      "loss": 0.9387,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484927205575985e-06,
+      "loss": 1.0048,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478766138100834e-06,
+      "loss": 0.9502,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472605661689863e-06,
+      "loss": 0.9321,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46644577873668e-06,
+      "loss": 1.0154,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460286491634664e-06,
+      "loss": 0.984,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454127802776961e-06,
+      "loss": 0.9075,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447969714556484e-06,
+      "loss": 0.9291,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441812229365913e-06,
+      "loss": 0.9534,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 0.9524,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42949907764403e-06,
+      "loss": 0.9435,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.423343415896904e-06,
+      "loss": 0.9337,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.417188366748051e-06,
+      "loss": 0.9954,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.411033932588969e-06,
+      "loss": 0.9668,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.404880115810914e-06,
+      "loss": 0.996,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.398726918804911e-06,
+      "loss": 0.9209,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.392574343961739e-06,
+      "loss": 0.9672,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.386422393671934e-06,
+      "loss": 0.903,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380271070325789e-06,
+      "loss": 0.8602,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374120376313362e-06,
+      "loss": 0.9178,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367970314024451e-06,
+      "loss": 0.9585,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361820885848623e-06,
+      "loss": 0.9028,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355672094175192e-06,
+      "loss": 0.9221,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349523941393224e-06,
+      "loss": 0.9626,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.343376429891538e-06,
+      "loss": 0.9843,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337229562058707e-06,
+      "loss": 0.9258,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331083340283051e-06,
+      "loss": 0.9615,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324937766952638e-06,
+      "loss": 0.9964,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318792844455282e-06,
+      "loss": 0.898,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312648575178552e-06,
+      "loss": 0.9288,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306504961509755e-06,
+      "loss": 0.9251,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300362005835947e-06,
+      "loss": 0.989,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294219710543931e-06,
+      "loss": 0.9503,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28807807802025e-06,
+      "loss": 0.9794,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281937110651187e-06,
+      "loss": 0.9703,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275796810822774e-06,
+      "loss": 0.9305,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269657180920773e-06,
+      "loss": 0.9941,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.9341,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25737994043779e-06,
+      "loss": 0.9768,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251242334627039e-06,
+      "loss": 0.9733,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245105408283162e-06,
+      "loss": 0.9724,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238969163790617e-06,
+      "loss": 0.9939,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232833603533601e-06,
+      "loss": 0.9666,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226698729896036e-06,
+      "loss": 0.9796,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.220564545261585e-06,
+      "loss": 0.9323,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214431052013636e-06,
+      "loss": 0.9661,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208298252535314e-06,
+      "loss": 0.9743,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202166149209475e-06,
+      "loss": 0.9381,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1960347444187e-06,
+      "loss": 0.9971,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189904040545302e-06,
+      "loss": 0.9276,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183774039971322e-06,
+      "loss": 0.9216,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177644745078525e-06,
+      "loss": 0.9787,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.171516158248406e-06,
+      "loss": 0.9963,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165388281862177e-06,
+      "loss": 0.9905,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159261118300784e-06,
+      "loss": 0.9505,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153134669944887e-06,
+      "loss": 1.0216,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147008939174875e-06,
+      "loss": 0.9958,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140883928370855e-06,
+      "loss": 0.8719,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134759639912655e-06,
+      "loss": 1.0164,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12863607617982e-06,
+      "loss": 0.9916,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12251323955162e-06,
+      "loss": 1.0064,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116391132407033e-06,
+      "loss": 0.9378,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11026975712476e-06,
+      "loss": 0.9966,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.104149116083216e-06,
+      "loss": 0.9278,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.098029211660534e-06,
+      "loss": 0.9881,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.091910046234552e-06,
+      "loss": 0.8459,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.085791622182832e-06,
+      "loss": 0.9265,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.079673941882639e-06,
+      "loss": 0.9213,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.073557007710957e-06,
+      "loss": 0.9835,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06744082204447e-06,
+      "loss": 0.9347,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061325387259578e-06,
+      "loss": 0.9484,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055210705732391e-06,
+      "loss": 0.9227,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.9611,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042983611954087e-06,
+      "loss": 0.9363,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036871204453718e-06,
+      "loss": 0.9429,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030759559712545e-06,
+      "loss": 1.0222,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0246486801052e-06,
+      "loss": 0.8247,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018538568006027e-06,
+      "loss": 0.9462,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012429225789055e-06,
+      "loss": 0.9315,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00632065582803e-06,
+      "loss": 0.9968,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00021286049639e-06,
+      "loss": 0.9204,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994105842167274e-06,
+      "loss": 0.9464,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987999603213518e-06,
+      "loss": 1.0189,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981894146007658e-06,
+      "loss": 0.981,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975789472921922e-06,
+      "loss": 1.0078,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96968558632824e-06,
+      "loss": 0.9392,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963582488598227e-06,
+      "loss": 0.9991,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957480182103198e-06,
+      "loss": 0.9687,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95137866921416e-06,
+      "loss": 1.0136,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945277952301811e-06,
+      "loss": 0.9893,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.939178033736538e-06,
+      "loss": 0.8831,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93307891588842e-06,
+      "loss": 0.8996,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926980601127225e-06,
+      "loss": 0.8893,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 0.9702,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914786390343112e-06,
+      "loss": 0.9741,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908690499058162e-06,
+      "loss": 0.9542,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902595420336074e-06,
+      "loss": 0.9135,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896501156545044e-06,
+      "loss": 0.9509,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890407710052957e-06,
+      "loss": 0.9677,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884315083227373e-06,
+      "loss": 0.9576,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878223278435539e-06,
+      "loss": 0.9899,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872132298044382e-06,
+      "loss": 1.0247,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866042144420502e-06,
+      "loss": 0.9153,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859952819930188e-06,
+      "loss": 0.9673,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853864326939403e-06,
+      "loss": 0.9356,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847776667813782e-06,
+      "loss": 0.988,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841689844918643e-06,
+      "loss": 1.0066,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835603860618973e-06,
+      "loss": 1.019,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829518717279439e-06,
+      "loss": 0.9257,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823434417264378e-06,
+      "loss": 0.9868,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817350962937802e-06,
+      "loss": 1.0049,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.811268356663387e-06,
+      "loss": 0.915,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805186600804489e-06,
+      "loss": 0.9429,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799105697724127e-06,
+      "loss": 0.9801,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793025649784991e-06,
+      "loss": 0.9485,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.786946459349441e-06,
+      "loss": 0.9426,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.780868128779499e-06,
+      "loss": 0.9507,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.9894,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.768714056682873e-06,
+      "loss": 0.9935,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.76263831987856e-06,
+      "loss": 0.9366,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.756563452384605e-06,
+      "loss": 0.8997,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750489456561351e-06,
+      "loss": 0.9568,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.744416334768809e-06,
+      "loss": 0.8778,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738344089366638e-06,
+      "loss": 0.8617,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73227272271417e-06,
+      "loss": 0.9522,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726202237170387e-06,
+      "loss": 0.9532,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720132635093935e-06,
+      "loss": 0.9946,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714063918843106e-06,
+      "loss": 0.9283,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70799609077586e-06,
+      "loss": 0.9514,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701929153249808e-06,
+      "loss": 0.9793,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695863108622208e-06,
+      "loss": 0.9109,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689797959249984e-06,
+      "loss": 0.9866,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6837337074897e-06,
+      "loss": 0.9763,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677670355697577e-06,
+      "loss": 0.9463,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.671607906229488e-06,
+      "loss": 0.9281,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66554636144095e-06,
+      "loss": 0.9588,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659485723687131e-06,
+      "loss": 0.8978,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653425995322852e-06,
+      "loss": 0.9148,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647367178702568e-06,
+      "loss": 0.9175,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641309276180394e-06,
+      "loss": 0.9505,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635252290110082e-06,
+      "loss": 0.951,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629196222845027e-06,
+      "loss": 0.9587,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623141076738271e-06,
+      "loss": 0.9547,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617086854142498e-06,
+      "loss": 0.9617,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611033557410027e-06,
+      "loss": 1.0217,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6049811888928235e-06,
+      "loss": 0.8823,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598929750942492e-06,
+      "loss": 0.9214,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592879245910273e-06,
+      "loss": 0.9458,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586829676147045e-06,
+      "loss": 0.9521,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580781044003324e-06,
+      "loss": 0.8978,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574733351829263e-06,
+      "loss": 0.9608,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5686866019746444e-06,
+      "loss": 0.8958,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562640796788893e-06,
+      "loss": 0.9489,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556595938621058e-06,
+      "loss": 0.9216,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550552029819828e-06,
+      "loss": 0.9266,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544509072733515e-06,
+      "loss": 0.9543,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53846706971007e-06,
+      "loss": 0.9647,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532426023097063e-06,
+      "loss": 0.9232,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5263859352417025e-06,
+      "loss": 0.9729,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5203468084908185e-06,
+      "loss": 0.9554,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.514308645190867e-06,
+      "loss": 0.9709,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508271447687936e-06,
+      "loss": 0.8972,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50223521832773e-06,
+      "loss": 0.9336,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.496199959455584e-06,
+      "loss": 0.9833,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.490165673416455e-06,
+      "loss": 0.954,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484132362554915e-06,
+      "loss": 0.9416,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478100029215166e-06,
+      "loss": 0.9398,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.472068675741024e-06,
+      "loss": 0.9332,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.466038304475931e-06,
+      "loss": 0.9238,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4600089177629384e-06,
+      "loss": 0.9528,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4539805179447234e-06,
+      "loss": 0.9244,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.447953107363574e-06,
+      "loss": 0.8908,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4419266883614e-06,
+      "loss": 0.9039,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435901263279717e-06,
+      "loss": 0.9894,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429876834459661e-06,
+      "loss": 0.9514,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4238534042419805e-06,
+      "loss": 0.9682,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.417830974967034e-06,
+      "loss": 0.9182,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.9457,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4057891286048356e-06,
+      "loss": 1.0086,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399769716196353e-06,
+      "loss": 0.9348,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.393751314088147e-06,
+      "loss": 0.9579,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387733924618617e-06,
+      "loss": 0.9443,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38171755012578e-06,
+      "loss": 0.9417,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3757021929472495e-06,
+      "loss": 1.0083,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3696878554202525e-06,
+      "loss": 0.9074,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3636745398816135e-06,
+      "loss": 1.0171,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35766224866776e-06,
+      "loss": 0.9884,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3516509841147276e-06,
+      "loss": 0.9702,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345640748558146e-06,
+      "loss": 1.0269,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33963154433325e-06,
+      "loss": 1.023,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333623373774868e-06,
+      "loss": 0.9796,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.327616239217432e-06,
+      "loss": 0.9536,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321610142994971e-06,
+      "loss": 0.9347,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315605087441107e-06,
+      "loss": 0.9301,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309601074889061e-06,
+      "loss": 0.9058,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303598107671648e-06,
+      "loss": 0.9608,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297596188121277e-06,
+      "loss": 0.9578,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291595318569951e-06,
+      "loss": 0.8917,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285595501349259e-06,
+      "loss": 0.9504,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279596738790389e-06,
+      "loss": 0.9713,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.273599033224114e-06,
+      "loss": 0.9394,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267602386980801e-06,
+      "loss": 0.9372,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2616068023904e-06,
+      "loss": 1.0292,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.255612281782451e-06,
+      "loss": 0.9312,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249618827486083e-06,
+      "loss": 0.9536,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243626441830009e-06,
+      "loss": 0.9105,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2376351271425235e-06,
+      "loss": 0.9135,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2316448857515076e-06,
+      "loss": 0.9508,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225655719984426e-06,
+      "loss": 0.9236,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219667632168326e-06,
+      "loss": 1.0011,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213680624629832e-06,
+      "loss": 0.9496,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207694699695155e-06,
+      "loss": 0.9957,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201709859690081e-06,
+      "loss": 0.9791,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1957261069399745e-06,
+      "loss": 0.9652,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.189743443769778e-06,
+      "loss": 0.9344,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183761872504011e-06,
+      "loss": 0.9805,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1777813954667716e-06,
+      "loss": 0.9338,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171802014981726e-06,
+      "loss": 1.022,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1658237333721216e-06,
+      "loss": 0.9139,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.159846552960774e-06,
+      "loss": 1.0098,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.153870476070073e-06,
+      "loss": 0.9468,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.14789550502198e-06,
+      "loss": 1.0144,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.141921642138025e-06,
+      "loss": 0.9361,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.135948889739307e-06,
+      "loss": 0.9605,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.129977250146496e-06,
+      "loss": 0.9293,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.124006725679828e-06,
+      "loss": 0.955,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118037318659108e-06,
+      "loss": 0.9403,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112069031403704e-06,
+      "loss": 0.9296,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106101866232546e-06,
+      "loss": 0.9149,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100135825464138e-06,
+      "loss": 1.0103,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094170911416543e-06,
+      "loss": 0.9513,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088207126407374e-06,
+      "loss": 0.9916,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082244472753823e-06,
+      "loss": 0.9938,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076282952772634e-06,
+      "loss": 0.8461,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070322568780111e-06,
+      "loss": 0.9891,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064363323092117e-06,
+      "loss": 1.0007,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058405218024074e-06,
+      "loss": 0.9794,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.052448255890958e-06,
+      "loss": 0.899,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046492439007306e-06,
+      "loss": 0.9884,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040537769687204e-06,
+      "loss": 0.9609,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.034584250244292e-06,
+      "loss": 0.8479,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028631882991771e-06,
+      "loss": 0.9353,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022680670242387e-06,
+      "loss": 0.9741,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.01673061430844e-06,
+      "loss": 0.9491,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.01078171750178e-06,
+      "loss": 1.0007,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004833982133808e-06,
+      "loss": 1.005,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.998887410515473e-06,
+      "loss": 0.9268,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992942004957271e-06,
+      "loss": 0.9484,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.986997767769243e-06,
+      "loss": 0.9729,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.981054701260981e-06,
+      "loss": 0.945,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97511280774162e-06,
+      "loss": 0.9468,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969172089519836e-06,
+      "loss": 1.0006,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963232548903853e-06,
+      "loss": 0.9351,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957294188201438e-06,
+      "loss": 0.9291,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951357009719893e-06,
+      "loss": 0.96,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9454210157660694e-06,
+      "loss": 0.9757,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939486208646349e-06,
+      "loss": 0.9491,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.933552590666659e-06,
+      "loss": 0.9831,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927620164132463e-06,
+      "loss": 0.9607,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921688931348762e-06,
+      "loss": 0.9153,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915758894620093e-06,
+      "loss": 0.9317,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.9094,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903902418543671e-06,
+      "loss": 1.0076,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8979759838026676e-06,
+      "loss": 0.9351,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892050754330181e-06,
+      "loss": 0.9867,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886126732428424e-06,
+      "loss": 0.9551,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8802039203991275e-06,
+      "loss": 0.9952,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874282320543557e-06,
+      "loss": 0.9965,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8683619351625065e-06,
+      "loss": 0.9181,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862442766556297e-06,
+      "loss": 0.9326,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85652481702478e-06,
+      "loss": 0.9316,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.850608088867332e-06,
+      "loss": 0.953,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.844692584382848e-06,
+      "loss": 0.9372,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.83877830586976e-06,
+      "loss": 0.9044,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.832865255626013e-06,
+      "loss": 0.8653,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.826953435949081e-06,
+      "loss": 0.965,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.821042849135957e-06,
+      "loss": 0.9866,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.815133497483157e-06,
+      "loss": 0.9385,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809225383286715e-06,
+      "loss": 0.9235,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8033185088421874e-06,
+      "loss": 0.9508,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797412876444645e-06,
+      "loss": 0.957,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791508488388675e-06,
+      "loss": 1.0035,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785605346968387e-06,
+      "loss": 0.9939,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779703454477402e-06,
+      "loss": 0.9701,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7738028132088565e-06,
+      "loss": 0.9134,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767903425455402e-06,
+      "loss": 0.9238,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762005293509202e-06,
+      "loss": 0.9626,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7561084196619306e-06,
+      "loss": 0.9849,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750212806204778e-06,
+      "loss": 0.9194,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744318455428436e-06,
+      "loss": 0.8882,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738425369623115e-06,
+      "loss": 0.9847,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732533551078528e-06,
+      "loss": 0.9262,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7266430020839e-06,
+      "loss": 0.9323,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720753724927957e-06,
+      "loss": 0.9603,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714865721898938e-06,
+      "loss": 0.9231,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708978995284581e-06,
+      "loss": 0.9239,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70309354737213e-06,
+      "loss": 0.9981,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.9253,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6913264967994375e-06,
+      "loss": 0.9868,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685444898711196e-06,
+      "loss": 0.8846,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679564588468861e-06,
+      "loss": 0.9473,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673685568357182e-06,
+      "loss": 0.9913,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66780784066041e-06,
+      "loss": 0.9642,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661931407662292e-06,
+      "loss": 1.0063,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656056271646076e-06,
+      "loss": 0.9697,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650182434894496e-06,
+      "loss": 0.9794,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644309899689794e-06,
+      "loss": 0.9355,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638438668313695e-06,
+      "loss": 0.9393,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632568743047426e-06,
+      "loss": 0.9445,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6267001261717015e-06,
+      "loss": 0.9179,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6208328199667305e-06,
+      "loss": 0.9038,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614966826712212e-06,
+      "loss": 1.0042,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609102148687333e-06,
+      "loss": 0.9583,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603238788170771e-06,
+      "loss": 0.9608,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597376747440694e-06,
+      "loss": 0.8951,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59151602877475e-06,
+      "loss": 0.8897,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5856566344500815e-06,
+      "loss": 0.9122,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.9398,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573941827930551e-06,
+      "loss": 0.9982,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568086420287389e-06,
+      "loss": 0.93,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562232346088903e-06,
+      "loss": 0.891,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5563796076096484e-06,
+      "loss": 0.8956,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550528207123667e-06,
+      "loss": 0.9595,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.544678146904475e-06,
+      "loss": 0.8087,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.538829429225068e-06,
+      "loss": 0.9461,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.532982056357928e-06,
+      "loss": 0.9827,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.527136030575005e-06,
+      "loss": 0.9778,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.521291354147727e-06,
+      "loss": 0.9527,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.515448029347005e-06,
+      "loss": 0.927,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5096060584432134e-06,
+      "loss": 0.9479,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503765443706212e-06,
+      "loss": 0.9654,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497926187405326e-06,
+      "loss": 0.9403,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492088291809355e-06,
+      "loss": 0.9425,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486251759186573e-06,
+      "loss": 0.9032,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48041659180472e-06,
+      "loss": 0.9414,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474582791931006e-06,
+      "loss": 0.915,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.468750361832112e-06,
+      "loss": 0.889,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462919303774186e-06,
+      "loss": 0.989,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4570896200228415e-06,
+      "loss": 0.8977,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45126131284316e-06,
+      "loss": 0.9665,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445434384499687e-06,
+      "loss": 0.9074,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439608837256432e-06,
+      "loss": 0.9528,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43378467337687e-06,
+      "loss": 0.9802,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427961895123935e-06,
+      "loss": 0.9001,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.422140504760025e-06,
+      "loss": 0.9874,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4163205045469975e-06,
+      "loss": 0.8435,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410501896746172e-06,
+      "loss": 0.8998,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.404684683618325e-06,
+      "loss": 0.9785,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398868867423691e-06,
+      "loss": 0.9251,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393054450421963e-06,
+      "loss": 0.8837,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387241434872293e-06,
+      "loss": 1.0078,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381429823033281e-06,
+      "loss": 0.9156,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375619617162985e-06,
+      "loss": 0.9101,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36981081951892e-06,
+      "loss": 0.9482,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364003432358051e-06,
+      "loss": 0.9323,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358197457936796e-06,
+      "loss": 0.9552,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3523928985110216e-06,
+      "loss": 0.9003,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 0.9247,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340788033666645e-06,
+      "loss": 0.9413,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334987732757028e-06,
+      "loss": 0.9682,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329188855860859e-06,
+      "loss": 0.9567,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32339140523125e-06,
+      "loss": 0.9312,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317595383120756e-06,
+      "loss": 0.9173,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311800791781381e-06,
+      "loss": 1.0044,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30600763346457e-06,
+      "loss": 0.947,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300215910421212e-06,
+      "loss": 0.9727,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294425624901638e-06,
+      "loss": 0.9677,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288636779155621e-06,
+      "loss": 0.9701,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282849375432374e-06,
+      "loss": 0.949,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277063415980549e-06,
+      "loss": 0.8663,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271278903048239e-06,
+      "loss": 0.9413,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265495838882973e-06,
+      "loss": 0.9702,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2597142257317185e-06,
+      "loss": 0.8916,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25393406584088e-06,
+      "loss": 0.8849,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248155361456293e-06,
+      "loss": 0.8995,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2423781148232345e-06,
+      "loss": 0.9209,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.236602328186405e-06,
+      "loss": 0.9034,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.230828003789949e-06,
+      "loss": 0.9094,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.225055143877434e-06,
+      "loss": 0.8735,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.219283750691864e-06,
+      "loss": 0.9721,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2135138264756715e-06,
+      "loss": 0.9948,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207745373470717e-06,
+      "loss": 0.914,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201978393918291e-06,
+      "loss": 0.9339,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196212890059111e-06,
+      "loss": 0.9342,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.190448864133321e-06,
+      "loss": 0.9192,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184686318380488e-06,
+      "loss": 0.9047,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17892525503961e-06,
+      "loss": 0.9447,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.8811,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1674075845468114e-06,
+      "loss": 0.921,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1616509818699975e-06,
+      "loss": 0.9748,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155895870555348e-06,
+      "loss": 0.8955,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150142252838967e-06,
+      "loss": 0.9302,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144390130956384e-06,
+      "loss": 0.9609,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138639507142539e-06,
+      "loss": 0.9395,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.132890383631796e-06,
+      "loss": 0.9481,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127142762657934e-06,
+      "loss": 0.9684,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12139664645415e-06,
+      "loss": 0.9845,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115652037253054e-06,
+      "loss": 0.9621,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109908937286673e-06,
+      "loss": 0.933,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104167348786443e-06,
+      "loss": 0.9743,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098427273983221e-06,
+      "loss": 1.024,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092688715107265e-06,
+      "loss": 0.9007,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.086951674388252e-06,
+      "loss": 0.9449,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0812161540552675e-06,
+      "loss": 0.9511,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075482156336804e-06,
+      "loss": 0.8791,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069749683460765e-06,
+      "loss": 0.9724,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06401873765446e-06,
+      "loss": 0.927,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058289321144608e-06,
+      "loss": 0.939,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052561436157329e-06,
+      "loss": 0.9215,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046835084918152e-06,
+      "loss": 0.9121,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041110269652005e-06,
+      "loss": 0.9502,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0353869925832255e-06,
+      "loss": 0.9141,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02966525593555e-06,
+      "loss": 0.9665,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023945061932119e-06,
+      "loss": 0.9443,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018226412795469e-06,
+      "loss": 0.9448,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0125093107475385e-06,
+      "loss": 0.9086,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00679375800967e-06,
+      "loss": 0.9573,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001079756802592e-06,
+      "loss": 0.9233,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9953673093464425e-06,
+      "loss": 0.9569,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98965641786075e-06,
+      "loss": 0.9436,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983947084564439e-06,
+      "loss": 0.9204,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978239311675826e-06,
+      "loss": 0.9197,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97253310141263e-06,
+      "loss": 0.9874,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966828455991951e-06,
+      "loss": 0.9271,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96112537763029e-06,
+      "loss": 0.9576,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955423868543537e-06,
+      "loss": 0.8568,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949723930946967e-06,
+      "loss": 0.9661,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.944025567055251e-06,
+      "loss": 0.9731,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9383287790824475e-06,
+      "loss": 0.972,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 0.9586,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.92693993974674e-06,
+      "loss": 0.9511,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.921247892808886e-06,
+      "loss": 0.9453,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.91555743064004e-06,
+      "loss": 0.9538,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909868555451191e-06,
+      "loss": 0.9022,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904181269452707e-06,
+      "loss": 0.9459,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898495574854342e-06,
+      "loss": 0.9714,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89281147386523e-06,
+      "loss": 0.913,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887128968693887e-06,
+      "loss": 0.9901,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881448061548208e-06,
+      "loss": 0.9781,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875768754635469e-06,
+      "loss": 0.9448,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870091050162321e-06,
+      "loss": 1.0167,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864414950334796e-06,
+      "loss": 0.9405,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858740457358298e-06,
+      "loss": 0.8801,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853067573437612e-06,
+      "loss": 0.9749,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847396300776893e-06,
+      "loss": 0.8854,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8417266415796745e-06,
+      "loss": 0.9642,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836058598048859e-06,
+      "loss": 0.915,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830392172386723e-06,
+      "loss": 0.9507,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824727366794915e-06,
+      "loss": 0.9385,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.819064183474451e-06,
+      "loss": 0.9272,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813402624625722e-06,
+      "loss": 0.9303,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807742692448483e-06,
+      "loss": 0.9203,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802084389141862e-06,
+      "loss": 1.031,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796427716904347e-06,
+      "loss": 0.9552,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790772677933799e-06,
+      "loss": 0.9549,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78511927442744e-06,
+      "loss": 0.9361,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77946750858186e-06,
+      "loss": 0.8998,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773817382593008e-06,
+      "loss": 1.0124,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7681688986562e-06,
+      "loss": 0.944,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7625220589661136e-06,
+      "loss": 0.931,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756876865716787e-06,
+      "loss": 0.9859,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751233321101617e-06,
+      "loss": 0.9886,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745591427313365e-06,
+      "loss": 1.0142,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739951186544141e-06,
+      "loss": 0.9731,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7343126009854215e-06,
+      "loss": 0.9583,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.728675672828037e-06,
+      "loss": 0.9085,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72304040426217e-06,
+      "loss": 0.9058,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717406797477371e-06,
+      "loss": 0.949,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711774854662535e-06,
+      "loss": 0.9243,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.706144578005908e-06,
+      "loss": 0.9216,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700515969695094e-06,
+      "loss": 0.9067,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694889031917047e-06,
+      "loss": 0.9972,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689263766858072e-06,
+      "loss": 0.9189,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683640176703824e-06,
+      "loss": 0.9567,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.678018263639309e-06,
+      "loss": 0.9476,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672398029848881e-06,
+      "loss": 0.9245,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666779477516234e-06,
+      "loss": 0.9729,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.9888,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65554742595583e-06,
+      "loss": 0.8908,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6499339310922015e-06,
+      "loss": 0.944,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6443221264146196e-06,
+      "loss": 0.8849,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.638712014103507e-06,
+      "loss": 0.9484,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.633103596338631e-06,
+      "loss": 0.9322,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.627496875299102e-06,
+      "loss": 0.936,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.62189185316337e-06,
+      "loss": 0.9505,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616288532109225e-06,
+      "loss": 0.9514,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610686914313796e-06,
+      "loss": 0.8976,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6050870019535496e-06,
+      "loss": 0.8535,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599488797204291e-06,
+      "loss": 0.9647,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5938923022411615e-06,
+      "loss": 0.9713,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588297519238637e-06,
+      "loss": 0.8947,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.582704450370533e-06,
+      "loss": 0.938,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5771130978099896e-06,
+      "loss": 0.8978,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571523463729487e-06,
+      "loss": 0.9548,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565935550300836e-06,
+      "loss": 0.949,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560349359695181e-06,
+      "loss": 0.8774,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554764894082994e-06,
+      "loss": 0.8615,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549182155634076e-06,
+      "loss": 1.0138,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543601146517563e-06,
+      "loss": 0.986,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5380218689019125e-06,
+      "loss": 1.0203,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532444324954911e-06,
+      "loss": 0.9538,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526868516843673e-06,
+      "loss": 0.9857,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521294446734637e-06,
+      "loss": 0.9334,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51572211679357e-06,
+      "loss": 0.9231,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510151529185554e-06,
+      "loss": 0.8891,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504582686075002e-06,
+      "loss": 0.9494,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499015589625649e-06,
+      "loss": 0.8955,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493450242000546e-06,
+      "loss": 0.9655,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4878866453620746e-06,
+      "loss": 0.9377,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4823248018719184e-06,
+      "loss": 0.9567,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4767647136910984e-06,
+      "loss": 0.8609,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471206382979941e-06,
+      "loss": 0.9953,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465649811898098e-06,
+      "loss": 0.9286,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.9515,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4545419572575265e-06,
+      "loss": 0.9382,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4489906780146716e-06,
+      "loss": 0.9205,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44344116703288e-06,
+      "loss": 0.9526,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43789342646837e-06,
+      "loss": 0.9549,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432347458476678e-06,
+      "loss": 0.8766,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4268032652126465e-06,
+      "loss": 0.991,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421260848830432e-06,
+      "loss": 0.9104,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415720211483499e-06,
+      "loss": 0.9454,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410181355324622e-06,
+      "loss": 0.9486,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404644282505882e-06,
+      "loss": 0.9816,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3991089951786725e-06,
+      "loss": 0.9807,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393575495493679e-06,
+      "loss": 0.8967,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388043785600907e-06,
+      "loss": 0.9473,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382513867649663e-06,
+      "loss": 0.888,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376985743788552e-06,
+      "loss": 0.8569,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37145941616549e-06,
+      "loss": 0.9151,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365934886927687e-06,
+      "loss": 0.964,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.360412158221661e-06,
+      "loss": 0.899,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.354891232193225e-06,
+      "loss": 0.9465,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.349372110987496e-06,
+      "loss": 0.9261,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.343854796748886e-06,
+      "loss": 0.9258,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.338339291621109e-06,
+      "loss": 0.9531,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.332825597747171e-06,
+      "loss": 0.9485,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32731371726938e-06,
+      "loss": 0.9594,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321803652329334e-06,
+      "loss": 0.879,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316295405067929e-06,
+      "loss": 0.9516,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3107889776253544e-06,
+      "loss": 0.9204,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305284372141095e-06,
+      "loss": 1.0002,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299781590753916e-06,
+      "loss": 0.9745,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294280635601889e-06,
+      "loss": 1.0003,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288781508822367e-06,
+      "loss": 0.9302,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283284212551997e-06,
+      "loss": 0.9495,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277788748926711e-06,
+      "loss": 0.9186,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2722951200817315e-06,
+      "loss": 0.9493,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2668033281515676e-06,
+      "loss": 0.9078,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2613133752700145e-06,
+      "loss": 0.9751,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2558252635701535e-06,
+      "loss": 0.9227,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.250338995184348e-06,
+      "loss": 0.9771,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244854572244249e-06,
+      "loss": 0.9402,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239371996880786e-06,
+      "loss": 0.9557,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2338912712241785e-06,
+      "loss": 0.8837,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228412397403916e-06,
+      "loss": 0.905,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222935377548777e-06,
+      "loss": 0.8355,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217460213786822e-06,
+      "loss": 1.0399,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211986908245374e-06,
+      "loss": 0.931,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206515463051054e-06,
+      "loss": 1.0021,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201045880329749e-06,
+      "loss": 0.9414,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195578162206627e-06,
+      "loss": 0.986,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190112310806126e-06,
+      "loss": 0.9029,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184648328251965e-06,
+      "loss": 0.9496,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179186216667133e-06,
+      "loss": 0.9445,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1737259781738934e-06,
+      "loss": 0.8492,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1682676148937805e-06,
+      "loss": 0.9634,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1628111289476025e-06,
+      "loss": 0.9514,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1573565224554365e-06,
+      "loss": 0.9308,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.151903797536631e-06,
+      "loss": 0.949,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146452956309798e-06,
+      "loss": 0.949,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.141004000892824e-06,
+      "loss": 1.0011,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135556933402862e-06,
+      "loss": 0.9534,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130111755956327e-06,
+      "loss": 0.8995,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1246684706689074e-06,
+      "loss": 0.9711,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119227079655544e-06,
+      "loss": 0.9095,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1137875850304545e-06,
+      "loss": 0.9217,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108349988907111e-06,
+      "loss": 1.0224,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.102914293398252e-06,
+      "loss": 0.9155,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097480500615879e-06,
+      "loss": 0.8917,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092048612671252e-06,
+      "loss": 0.8999,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086618631674888e-06,
+      "loss": 0.9168,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081190559736569e-06,
+      "loss": 0.9166,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.075764398965331e-06,
+      "loss": 1.013,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.070340151469468e-06,
+      "loss": 0.9634,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.064917819356532e-06,
+      "loss": 0.9482,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.059497404733331e-06,
+      "loss": 0.9599,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.054078909705926e-06,
+      "loss": 0.8899,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.048662336379635e-06,
+      "loss": 1.0175,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043247686859024e-06,
+      "loss": 0.8698,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037834963247922e-06,
+      "loss": 0.9366,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032424167649394e-06,
+      "loss": 0.9298,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027015302165768e-06,
+      "loss": 0.9157,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021608368898621e-06,
+      "loss": 0.9781,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016203369948774e-06,
+      "loss": 0.896,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0108003074163005e-06,
+      "loss": 0.9352,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005399183400521e-06,
+      "loss": 0.8991,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.9764,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994602759312557e-06,
+      "loss": 0.9599,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989207463435244e-06,
+      "loss": 0.9159,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9838141144643636e-06,
+      "loss": 0.9501,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978422714495465e-06,
+      "loss": 0.9004,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973033265623333e-06,
+      "loss": 0.9674,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967645769942e-06,
+      "loss": 0.8982,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962260229544738e-06,
+      "loss": 0.917,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956876646524059e-06,
+      "loss": 0.9445,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951495022971713e-06,
+      "loss": 0.9676,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946115360978696e-06,
+      "loss": 0.9889,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940737662635227e-06,
+      "loss": 0.9387,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935361930030774e-06,
+      "loss": 0.8793,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929988165254041e-06,
+      "loss": 0.9719,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924616370392962e-06,
+      "loss": 0.8946,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919246547534709e-06,
+      "loss": 1.0019,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9138786987656865e-06,
+      "loss": 1.0506,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908512826171533e-06,
+      "loss": 0.9401,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9031489318371195e-06,
+      "loss": 0.8969,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897787017846547e-06,
+      "loss": 0.9461,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892427086283147e-06,
+      "loss": 0.8522,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887069139229481e-06,
+      "loss": 0.9237,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881713178767341e-06,
+      "loss": 1.0092,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8763592069777454e-06,
+      "loss": 0.9648,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871007225940939e-06,
+      "loss": 0.9788,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865657237736397e-06,
+      "loss": 0.9521,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860309244442821e-06,
+      "loss": 1.0092,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854963248138121e-06,
+      "loss": 1.0471,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849619250899458e-06,
+      "loss": 0.9352,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844277254803198e-06,
+      "loss": 0.8962,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844277254803198e-06,
+      "loss": 0.905,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838937261924933e-06,
+      "loss": 0.8769,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83359927433948e-06,
+      "loss": 0.8915,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8282632941208725e-06,
+      "loss": 0.9134,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822929323342369e-06,
+      "loss": 0.9514,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817597364076441e-06,
+      "loss": 0.9251,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812267418394784e-06,
+      "loss": 0.8998,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806939488368308e-06,
+      "loss": 0.9001,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8016135760671435e-06,
+      "loss": 0.9858,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.796289683560631e-06,
+      "loss": 0.9959,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.790967812917334e-06,
+      "loss": 0.918,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.78564796620502e-06,
+      "loss": 0.9205,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7803301454906804e-06,
+      "loss": 0.9627,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.775014352840512e-06,
+      "loss": 0.9455,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76970059031993e-06,
+      "loss": 0.9082,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764388859993556e-06,
+      "loss": 0.8976,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.759079163925223e-06,
+      "loss": 0.9333,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753771504177976e-06,
+      "loss": 1.0364,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7484658828140665e-06,
+      "loss": 0.9444,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743162301894952e-06,
+      "loss": 0.9831,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7378607634813045e-06,
+      "loss": 0.9428,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732561269632992e-06,
+      "loss": 0.9504,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727263822409099e-06,
+      "loss": 0.965,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7219684238679066e-06,
+      "loss": 0.9551,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716675076066902e-06,
+      "loss": 0.9336,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711383781062779e-06,
+      "loss": 0.917,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706094540911429e-06,
+      "loss": 0.9568,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700807357667953e-06,
+      "loss": 1.0108,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6955222333866365e-06,
+      "loss": 0.8581,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6902391701209826e-06,
+      "loss": 0.9231,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684958169923684e-06,
+      "loss": 0.9351,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.8874,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674402366940927e-06,
+      "loss": 0.9173,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669127568256851e-06,
+      "loss": 0.9008,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663854840843885e-06,
+      "loss": 1.0135,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658584186750713e-06,
+      "loss": 0.9655,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6533156080252076e-06,
+      "loss": 0.9203,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648049106714434e-06,
+      "loss": 0.8786,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642784684864656e-06,
+      "loss": 0.9175,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637522344521323e-06,
+      "loss": 0.9939,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63226208772908e-06,
+      "loss": 0.8976,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627003916531761e-06,
+      "loss": 0.9348,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621747832972389e-06,
+      "loss": 0.9884,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616493839093179e-06,
+      "loss": 0.961,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6112419369355255e-06,
+      "loss": 0.9717,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60599212854002e-06,
+      "loss": 0.9805,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600744415946438e-06,
+      "loss": 0.9042,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600744415946438e-06,
+      "loss": 0.8967,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595498801193736e-06,
+      "loss": 0.9354,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590255286320062e-06,
+      "loss": 0.9881,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5850138733627435e-06,
+      "loss": 0.9687,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5797745643582935e-06,
+      "loss": 0.9856,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5745373613424075e-06,
+      "loss": 0.9429,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5693022663499584e-06,
+      "loss": 0.9802,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564069281415009e-06,
+      "loss": 0.9606,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5588384085707925e-06,
+      "loss": 1.0001,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5536096498497295e-06,
+      "loss": 0.9473,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548383007283412e-06,
+      "loss": 0.9558,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543158482902617e-06,
+      "loss": 0.9221,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537936078737291e-06,
+      "loss": 0.9623,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.532715796816565e-06,
+      "loss": 0.9596,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.52749763916874e-06,
+      "loss": 0.9208,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.522281607821288e-06,
+      "loss": 0.9465,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.517067704800864e-06,
+      "loss": 0.9895,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.511855932133289e-06,
+      "loss": 0.939,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.506646291843561e-06,
+      "loss": 0.9035,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.501438785955845e-06,
+      "loss": 0.9552,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.496233416493481e-06,
+      "loss": 0.9493,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491030185478976e-06,
+      "loss": 0.9368,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4858290949340065e-06,
+      "loss": 0.9569,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480630146879419e-06,
+      "loss": 0.923,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475433343335225e-06,
+      "loss": 0.9747,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470238686320606e-06,
+      "loss": 0.9492,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4650461778539055e-06,
+      "loss": 0.9527,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459855819952635e-06,
+      "loss": 0.9842,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454667614633471e-06,
+      "loss": 0.9695,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449481563912252e-06,
+      "loss": 0.9501,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.9168,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439115934322816e-06,
+      "loss": 0.9661,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433936359482087e-06,
+      "loss": 0.9043,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428758947294278e-06,
+      "loss": 0.9444,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423583699771036e-06,
+      "loss": 0.9579,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418410618923163e-06,
+      "loss": 0.9613,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413239706760621e-06,
+      "loss": 1.0166,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408070965292534e-06,
+      "loss": 0.9184,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402904396527172e-06,
+      "loss": 0.9397,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397740002471973e-06,
+      "loss": 0.9105,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392577785133521e-06,
+      "loss": 0.9677,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387417746517557e-06,
+      "loss": 0.9377,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382259888628978e-06,
+      "loss": 0.9243,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377104213471831e-06,
+      "loss": 0.9568,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371950723049314e-06,
+      "loss": 0.9338,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3667994193637794e-06,
+      "loss": 0.8946,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361650304416728e-06,
+      "loss": 0.9584,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356503380208812e-06,
+      "loss": 0.9943,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351358648739826e-06,
+      "loss": 0.9234,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34621611200872e-06,
+      "loss": 0.991,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3410757720135886e-06,
+      "loss": 0.9601,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335937630751675e-06,
+      "loss": 0.9687,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330801690219361e-06,
+      "loss": 0.984,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325667952412184e-06,
+      "loss": 0.9313,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320536419324816e-06,
+      "loss": 0.9636,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315407092951078e-06,
+      "loss": 0.9601,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310279975283932e-06,
+      "loss": 0.9588,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305155068315481e-06,
+      "loss": 0.8452,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30003237403697e-06,
+      "loss": 0.9787,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2949118944387846e-06,
+      "loss": 0.9446,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289793631510449e-06,
+      "loss": 0.9101,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284677587240625e-06,
+      "loss": 0.8847,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2795637636171165e-06,
+      "loss": 0.9308,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.27445216262686e-06,
+      "loss": 0.977,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2693427862559335e-06,
+      "loss": 0.9692,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.264235636489542e-06,
+      "loss": 0.9495,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2591307153120335e-06,
+      "loss": 0.8987,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.254028024706887e-06,
+      "loss": 0.8997,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.248927566656717e-06,
+      "loss": 0.9867,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2438293431432665e-06,
+      "loss": 0.9099,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.238733356147414e-06,
+      "loss": 0.9139,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.233639607649168e-06,
+      "loss": 0.9314,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.228548099627665e-06,
+      "loss": 0.8962,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223458834061175e-06,
+      "loss": 0.9124,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218371812927092e-06,
+      "loss": 0.965,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213287038201943e-06,
+      "loss": 0.9678,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208204511861379e-06,
+      "loss": 0.9309,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203124235880179e-06,
+      "loss": 0.9486,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198046212232243e-06,
+      "loss": 0.9106,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192970442890602e-06,
+      "loss": 0.9473,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187896929827414e-06,
+      "loss": 0.948,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182825675013945e-06,
+      "loss": 0.9761,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1777566804206e-06,
+      "loss": 0.9355,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172689948016896e-06,
+      "loss": 1.0162,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167625479771476e-06,
+      "loss": 0.8421,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162563277652104e-06,
+      "loss": 0.9023,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157503343625659e-06,
+      "loss": 0.9746,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152445679658139e-06,
+      "loss": 0.8361,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147390287714667e-06,
+      "loss": 0.9877,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142337169759472e-06,
+      "loss": 0.9588,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137286327755913e-06,
+      "loss": 0.8555,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132237763666451e-06,
+      "loss": 0.9681,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1271914794526715e-06,
+      "loss": 0.9736,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.9674,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117105758494056e-06,
+      "loss": 0.9697,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112066325667954e-06,
+      "loss": 0.9782,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107029180554996e-06,
+      "loss": 0.9491,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101994325112332e-06,
+      "loss": 0.9487,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096961761296211e-06,
+      "loss": 0.9552,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0919314910620025e-06,
+      "loss": 0.9442,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.086903516364179e-06,
+      "loss": 0.9257,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0818778391563255e-06,
+      "loss": 0.9922,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07685446139113e-06,
+      "loss": 0.9965,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07183338502039e-06,
+      "loss": 0.9695,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066814611995006e-06,
+      "loss": 0.9778,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061798144264986e-06,
+      "loss": 0.9073,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056783983779442e-06,
+      "loss": 0.953,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051772132486589e-06,
+      "loss": 0.8929,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046762592333744e-06,
+      "loss": 0.9272,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041755365267323e-06,
+      "loss": 0.9672,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03675045323286e-06,
+      "loss": 0.8714,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03675045323286e-06,
+      "loss": 0.888,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0317478581749644e-06,
+      "loss": 0.9601,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02674758203736e-06,
+      "loss": 0.9168,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021749626762869e-06,
+      "loss": 0.9149,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01675399429341e-06,
+      "loss": 0.931,
+      "step": 3714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0117606865699975e-06,
+      "loss": 0.9058,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.006769705532746e-06,
+      "loss": 0.9642,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.001781053120863e-06,
+      "loss": 0.8904,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.996794731272653e-06,
+      "loss": 0.9375,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.991810741925515e-06,
+      "loss": 0.9343,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.986829087015941e-06,
+      "loss": 0.966,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981849768479516e-06,
+      "loss": 0.9618,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.976872788250918e-06,
+      "loss": 0.9138,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.971898148263915e-06,
+      "loss": 0.9713,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.966925850451367e-06,
+      "loss": 0.9343,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.961955896745224e-06,
+      "loss": 0.9476,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956988289076529e-06,
+      "loss": 0.933,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9520230293754e-06,
+      "loss": 0.8989,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9470601195710575e-06,
+      "loss": 0.9257,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942099561591802e-06,
+      "loss": 0.9367,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937141357365023e-06,
+      "loss": 0.8942,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9321855088171934e-06,
+      "loss": 0.9044,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927232017873872e-06,
+      "loss": 0.9802,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922280886459701e-06,
+      "loss": 0.9719,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917332116498406e-06,
+      "loss": 1.0009,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912385709912794e-06,
+      "loss": 0.9678,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9074416686247554e-06,
+      "loss": 0.968,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902499994555261e-06,
+      "loss": 0.9144,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897560689624362e-06,
+      "loss": 0.8754,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892623755751189e-06,
+      "loss": 0.9143,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887689194853951e-06,
+      "loss": 0.9861,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882757008849936e-06,
+      "loss": 0.9385,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877827199655506e-06,
+      "loss": 0.9317,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872899769186107e-06,
+      "loss": 0.9421,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867974719356249e-06,
+      "loss": 0.9087,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8630520520795275e-06,
+      "loss": 0.9153,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858131769268607e-06,
+      "loss": 0.922,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853213872835229e-06,
+      "loss": 0.9364,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848298364690204e-06,
+      "loss": 0.908,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 0.9449,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838474520903825e-06,
+      "loss": 0.9331,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833566189079453e-06,
+      "loss": 0.9476,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8286602531773975e-06,
+      "loss": 0.957,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823756715103822e-06,
+      "loss": 0.986,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8188555767639625e-06,
+      "loss": 0.959,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8139568400621184e-06,
+      "loss": 0.9167,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8090605069016596e-06,
+      "loss": 0.9328,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804166579185018e-06,
+      "loss": 0.8847,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799275058813694e-06,
+      "loss": 0.921,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794385947688255e-06,
+      "loss": 0.9154,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7894992477083226e-06,
+      "loss": 0.8752,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78461496077259e-06,
+      "loss": 0.9435,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779733088778813e-06,
+      "loss": 0.9817,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774853633623806e-06,
+      "loss": 0.9849,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7699765972034442e-06,
+      "loss": 0.9303,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.9813,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7602297881454653e-06,
+      "loss": 0.9269,
+      "step": 3766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.755360019294898e-06,
+      "loss": 0.9696,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.750492676753077e-06,
+      "loss": 0.9342,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7456277624111725e-06,
+      "loss": 0.9029,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7407652781594094e-06,
+      "loss": 0.9584,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.735905225887072e-06,
+      "loss": 0.9055,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7310476074824966e-06,
+      "loss": 0.9217,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.726192424833074e-06,
+      "loss": 0.8567,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7213396798252522e-06,
+      "loss": 0.9516,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7164893743445274e-06,
+      "loss": 0.9654,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7116415102754545e-06,
+      "loss": 0.8628,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7067960895016277e-06,
+      "loss": 0.9321,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701953113905703e-06,
+      "loss": 0.976,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6971125853693834e-06,
+      "loss": 0.9397,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692274505773419e-06,
+      "loss": 0.945,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.687438876997612e-06,
+      "loss": 0.914,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6826057009208083e-06,
+      "loss": 0.8879,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677774979420904e-06,
+      "loss": 0.9154,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.67294671437484e-06,
+      "loss": 0.8874,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6681209076586035e-06,
+      "loss": 0.9293,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.663297561147224e-06,
+      "loss": 0.9732,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6584766767147796e-06,
+      "loss": 0.8813,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6536582562343893e-06,
+      "loss": 0.9588,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6488423015782128e-06,
+      "loss": 0.9254,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.644028814617454e-06,
+      "loss": 0.9399,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6392177972223596e-06,
+      "loss": 0.9049,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6344092512622173e-06,
+      "loss": 0.9551,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6296031786053455e-06,
+      "loss": 0.9703,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6247995811191107e-06,
+      "loss": 0.9783,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619998460669916e-06,
+      "loss": 0.8729,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6151998191232005e-06,
+      "loss": 0.9084,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.610403658343443e-06,
+      "loss": 0.9483,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6056099801941535e-06,
+      "loss": 0.88,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.600818786537883e-06,
+      "loss": 0.8877,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5960300792362124e-06,
+      "loss": 0.9545,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591243860149759e-06,
+      "loss": 0.9304,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.586460131138174e-06,
+      "loss": 0.9001,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5816788940601387e-06,
+      "loss": 0.9781,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.576900150773368e-06,
+      "loss": 0.9667,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.9167,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5673501529996314e-06,
+      "loss": 0.8859,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5625789022232473e-06,
+      "loss": 0.8241,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5578101526592877e-06,
+      "loss": 0.9199,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5530439061606202e-06,
+      "loss": 0.9865,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5482801645791266e-06,
+      "loss": 0.9146,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5435189297657267e-06,
+      "loss": 0.9408,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5387602035703637e-06,
+      "loss": 0.9633,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534003987842005e-06,
+      "loss": 0.8904,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5292502844286447e-06,
+      "loss": 0.9264,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5244990951772972e-06,
+      "loss": 0.8219,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5197504219340027e-06,
+      "loss": 0.9346,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5150042665438233e-06,
+      "loss": 0.9881,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5102606308508423e-06,
+      "loss": 0.8992,
+      "step": 3818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.505519516698165e-06,
+      "loss": 0.9016,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5007809259279146e-06,
+      "loss": 0.8552,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.496044860381238e-06,
+      "loss": 0.9667,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4913113218982965e-06,
+      "loss": 0.9788,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.486580312318273e-06,
+      "loss": 0.9249,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.481851833479365e-06,
+      "loss": 0.9963,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4771258872187917e-06,
+      "loss": 0.991,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.472402475372778e-06,
+      "loss": 0.9262,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4676815997765756e-06,
+      "loss": 0.8695,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4629632622644437e-06,
+      "loss": 0.8349,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4582474646696575e-06,
+      "loss": 1.001,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453534208824507e-06,
+      "loss": 0.9727,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448823496560292e-06,
+      "loss": 0.929,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444115329707325e-06,
+      "loss": 0.9082,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4394097100949286e-06,
+      "loss": 0.8855,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4347066395514383e-06,
+      "loss": 1.0114,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4300061199041967e-06,
+      "loss": 0.936,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.425308152979555e-06,
+      "loss": 0.9522,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4206127406028744e-06,
+      "loss": 0.9492,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415919884598523e-06,
+      "loss": 0.9136,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411229586789876e-06,
+      "loss": 1.0005,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4065418489993118e-06,
+      "loss": 0.9125,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.401856673048217e-06,
+      "loss": 0.9097,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3971740607569868e-06,
+      "loss": 0.9418,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.392494013945009e-06,
+      "loss": 0.9401,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3878165344306824e-06,
+      "loss": 0.9125,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3831416240314085e-06,
+      "loss": 0.9307,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.378469284563589e-06,
+      "loss": 1.0365,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3737995178426276e-06,
+      "loss": 0.9373,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3691323256829266e-06,
+      "loss": 0.9189,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3644677098978894e-06,
+      "loss": 0.977,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3598056722999185e-06,
+      "loss": 0.8877,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3551462147004132e-06,
+      "loss": 0.9351,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3504893389097727e-06,
+      "loss": 0.9179,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3458350467373914e-06,
+      "loss": 0.9318,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3411833399916584e-06,
+      "loss": 0.974,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3365342204799613e-06,
+      "loss": 0.9624,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3318876900086815e-06,
+      "loss": 0.8851,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3272437503831945e-06,
+      "loss": 0.9572,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3226024034078698e-06,
+      "loss": 0.8888,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3179636508860645e-06,
+      "loss": 0.9447,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3133274946201333e-06,
+      "loss": 0.8916,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.9571,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304062978060263e-06,
+      "loss": 0.9689,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2994346213659824e-06,
+      "loss": 0.8901,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.294808868126894e-06,
+      "loss": 0.9373,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290185720140301e-06,
+      "loss": 0.9878,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285565179202491e-06,
+      "loss": 0.956,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280947247108741e-06,
+      "loss": 0.9324,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2763319256533177e-06,
+      "loss": 0.978,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2717192166294685e-06,
+      "loss": 0.9283,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267109121829428e-06,
+      "loss": 0.9681,
+      "step": 3870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262501643044417e-06,
+      "loss": 0.975,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2578967820646313e-06,
+      "loss": 0.8987,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2532945406792573e-06,
+      "loss": 0.9643,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.248694920676464e-06,
+      "loss": 0.9335,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2440979238433977e-06,
+      "loss": 0.8697,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.23950355196619e-06,
+      "loss": 0.9593,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.234911806829948e-06,
+      "loss": 0.9806,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2303226902187625e-06,
+      "loss": 0.922,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.225736203915698e-06,
+      "loss": 0.966,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.221152349702802e-06,
+      "loss": 0.9087,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216571129361097e-06,
+      "loss": 0.9409,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2119925446705824e-06,
+      "loss": 0.9144,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2074165974102324e-06,
+      "loss": 0.9563,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2028432893579975e-06,
+      "loss": 0.9431,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1982726222908046e-06,
+      "loss": 0.9538,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1937045979845518e-06,
+      "loss": 0.9206,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1891392182141103e-06,
+      "loss": 0.968,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1845764847533276e-06,
+      "loss": 0.8821,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1800163993750166e-06,
+      "loss": 0.9302,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1754589638509647e-06,
+      "loss": 0.9646,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1709041799519312e-06,
+      "loss": 0.9075,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166352049447644e-06,
+      "loss": 0.9338,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161802574106799e-06,
+      "loss": 0.9048,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157255755697062e-06,
+      "loss": 0.9316,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152711595985065e-06,
+      "loss": 0.9226,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148170096736408e-06,
+      "loss": 1.0547,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143631259715658e-06,
+      "loss": 0.9468,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1390950866863456e-06,
+      "loss": 0.9806,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.134561579410966e-06,
+      "loss": 1.0337,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1300307396509833e-06,
+      "loss": 0.9401,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1255025691668184e-06,
+      "loss": 0.8826,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.120977069717861e-06,
+      "loss": 0.8809,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116454243062459e-06,
+      "loss": 0.9342,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1119340909579265e-06,
+      "loss": 0.9021,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10741661516053e-06,
+      "loss": 0.8874,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.102901817425504e-06,
+      "loss": 0.9803,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0983896995070397e-06,
+      "loss": 0.9438,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0938802631582865e-06,
+      "loss": 0.9407,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089373510131354e-06,
+      "loss": 0.9503,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0848694421773075e-06,
+      "loss": 0.9394,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0803680610461674e-06,
+      "loss": 0.9603,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0758693684869147e-06,
+      "loss": 0.947,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.071373366247482e-06,
+      "loss": 0.9824,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.066880056074758e-06,
+      "loss": 0.9454,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0623894397145837e-06,
+      "loss": 0.9299,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.057901518911758e-06,
+      "loss": 0.856,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0534162954100264e-06,
+      "loss": 0.9359,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0489337709520918e-06,
+      "loss": 0.9462,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.044453947279603e-06,
+      "loss": 0.963,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0399768261331664e-06,
+      "loss": 0.9237,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0355024092523334e-06,
+      "loss": 0.9439,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031030698375602e-06,
+      "loss": 0.9391,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0265616952404252e-06,
+      "loss": 0.8979,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0220954015832004e-06,
+      "loss": 0.9419,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.017631819139273e-06,
+      "loss": 0.9178,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.013170949642935e-06,
+      "loss": 0.9615,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0087127948274264e-06,
+      "loss": 0.9755,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.004257356424927e-06,
+      "loss": 0.9313,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.999804636166567e-06,
+      "loss": 0.9827,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.995354635782417e-06,
+      "loss": 0.9817,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.990907357001491e-06,
+      "loss": 0.9627,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.986462801551748e-06,
+      "loss": 0.8951,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9820209711600858e-06,
+      "loss": 1.0063,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977581867552346e-06,
+      "loss": 0.9386,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9731454924533086e-06,
+      "loss": 0.9592,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9687118475866948e-06,
+      "loss": 0.9528,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9642809346751677e-06,
+      "loss": 0.8658,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9598527554403187e-06,
+      "loss": 0.8691,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9554273116026887e-06,
+      "loss": 0.9449,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.95100460488175e-06,
+      "loss": 0.9998,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9465846369959126e-06,
+      "loss": 0.9464,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942167409662524e-06,
+      "loss": 0.9699,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.937752924597864e-06,
+      "loss": 0.9218,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933341183517148e-06,
+      "loss": 0.9374,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.9916,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.92452594016308e-06,
+      "loss": 0.8904,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9201224413148254e-06,
+      "loss": 0.8635,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9157216933007095e-06,
+      "loss": 0.9158,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.91132369783061e-06,
+      "loss": 0.9298,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.906928456613336e-06,
+      "loss": 0.9553,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.902535971356625e-06,
+      "loss": 0.9185,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.898146243767146e-06,
+      "loss": 0.9532,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.893759275550494e-06,
+      "loss": 0.9617,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8893750684111977e-06,
+      "loss": 0.9631,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884993624052701e-06,
+      "loss": 0.9077,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8806149441773844e-06,
+      "loss": 0.9177,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876239030486554e-06,
+      "loss": 0.9092,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.871865884680436e-06,
+      "loss": 0.8881,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.867495508458186e-06,
+      "loss": 0.904,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8631279035178796e-06,
+      "loss": 0.9376,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8587630715565185e-06,
+      "loss": 0.9276,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8544010142700263e-06,
+      "loss": 0.9874,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850041733353247e-06,
+      "loss": 0.9528,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845685230499948e-06,
+      "loss": 0.8339,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8413315074028157e-06,
+      "loss": 0.951,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8369805657534576e-06,
+      "loss": 0.8096,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8326324072423994e-06,
+      "loss": 0.903,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828287033559086e-06,
+      "loss": 0.9868,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.823944446391881e-06,
+      "loss": 0.9478,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.819604647428067e-06,
+      "loss": 0.9506,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8152676383538346e-06,
+      "loss": 0.9258,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8109334208542994e-06,
+      "loss": 0.9926,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8066019966134907e-06,
+      "loss": 0.9002,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8022733673143508e-06,
+      "loss": 0.9118,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7979475346387363e-06,
+      "loss": 0.9299,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.793624500267418e-06,
+      "loss": 0.9256,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7893042658800793e-06,
+      "loss": 0.9356,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.784986833155314e-06,
+      "loss": 0.9385,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.78067220377063e-06,
+      "loss": 0.9416,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.776360379402445e-06,
+      "loss": 0.9111,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7720513617260857e-06,
+      "loss": 0.898,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7677451524157882e-06,
+      "loss": 0.9285,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.763441753144701e-06,
+      "loss": 1.011,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7591411655848755e-06,
+      "loss": 0.9517,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7548433914072736e-06,
+      "loss": 0.9474,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7505484322817656e-06,
+      "loss": 0.9411,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746256289877126e-06,
+      "loss": 0.9421,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.741966965861033e-06,
+      "loss": 0.8834,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7376804619000706e-06,
+      "loss": 0.9181,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7333967796597317e-06,
+      "loss": 0.844,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729115920804408e-06,
+      "loss": 0.9141,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7248378869973946e-06,
+      "loss": 0.8766,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.720562679900892e-06,
+      "loss": 0.9528,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716290301175999e-06,
+      "loss": 0.9646,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712020752482717e-06,
+      "loss": 0.8508,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707754035479948e-06,
+      "loss": 0.8769,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.703490151825492e-06,
+      "loss": 0.9675,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.699229103176052e-06,
+      "loss": 0.903,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.694970891187225e-06,
+      "loss": 0.9076,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.690715517513508e-06,
+      "loss": 0.9748,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6864629838082957e-06,
+      "loss": 0.9424,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6822132917238785e-06,
+      "loss": 0.9876,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6779664429114448e-06,
+      "loss": 0.8923,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6737224390210726e-06,
+      "loss": 0.917,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 0.908,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6652429726013162e-06,
+      "loss": 0.8966,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661007513366567e-06,
+      "loss": 1.0153,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.656774905643147e-06,
+      "loss": 0.9949,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652545151075606e-06,
+      "loss": 0.9507,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6483182513073835e-06,
+      "loss": 0.9187,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6440942079808096e-06,
+      "loss": 0.9472,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.639873022737105e-06,
+      "loss": 0.9289,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.635654697216382e-06,
+      "loss": 0.8933,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.631439233057638e-06,
+      "loss": 0.9834,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6272266318987606e-06,
+      "loss": 0.9863,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6230168953765257e-06,
+      "loss": 0.9584,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6188100251265947e-06,
+      "loss": 0.8774,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6146060227835156e-06,
+      "loss": 0.8436,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6104048899807243e-06,
+      "loss": 0.886,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6062066283505404e-06,
+      "loss": 0.9711,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6020112395241627e-06,
+      "loss": 0.9092,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5978187251316823e-06,
+      "loss": 0.9339,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5936290868020685e-06,
+      "loss": 0.9784,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5894423261631717e-06,
+      "loss": 0.895,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5852584448417327e-06,
+      "loss": 0.9021,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.581077444463365e-06,
+      "loss": 0.9269,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5768993266525653e-06,
+      "loss": 0.9574,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5727240930327103e-06,
+      "loss": 0.8951,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 0.9508,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.564382284853738e-06,
+      "loss": 0.8521,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5602157135357697e-06,
+      "loss": 0.9199,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5560520328910397e-06,
+      "loss": 0.9375,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.55189124453732e-06,
+      "loss": 0.9621,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5477333500912473e-06,
+      "loss": 0.8655,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5435783511683444e-06,
+      "loss": 0.9435,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.539426249383006e-06,
+      "loss": 0.8671,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5352770463484986e-06,
+      "loss": 0.9005,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531130743676966e-06,
+      "loss": 0.9264,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5269873429794235e-06,
+      "loss": 0.9417,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5228468458657585e-06,
+      "loss": 0.9158,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.51870925394473e-06,
+      "loss": 0.9531,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5145745688239696e-06,
+      "loss": 0.859,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5104427921099783e-06,
+      "loss": 0.928,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506313925408127e-06,
+      "loss": 0.9786,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502187970322657e-06,
+      "loss": 0.887,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4980649284566783e-06,
+      "loss": 0.883,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493944801412167e-06,
+      "loss": 0.9719,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489827590789968e-06,
+      "loss": 0.9587,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485713298189798e-06,
+      "loss": 0.932,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4816019252102274e-06,
+      "loss": 0.8757,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4774934734487034e-06,
+      "loss": 0.9517,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.473387944501533e-06,
+      "loss": 0.9508,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469285339963892e-06,
+      "loss": 0.9912,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4651856614298143e-06,
+      "loss": 0.9487,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461088910492202e-06,
+      "loss": 0.7636,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.456995088742815e-06,
+      "loss": 0.9496,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45290419777228e-06,
+      "loss": 0.9464,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.448816239170081e-06,
+      "loss": 0.917,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444731214524565e-06,
+      "loss": 0.9264,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440649125422937e-06,
+      "loss": 0.9699,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436569973451264e-06,
+      "loss": 0.9206,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.432493760194469e-06,
+      "loss": 0.8744,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428420487236336e-06,
+      "loss": 0.9403,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4243501561595027e-06,
+      "loss": 0.9382,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420282768545469e-06,
+      "loss": 0.9865,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.41621832597459e-06,
+      "loss": 0.8613,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412156830026069e-06,
+      "loss": 1.0094,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408098282277972e-06,
+      "loss": 0.9632,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4040426843072206e-06,
+      "loss": 0.8052,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3999900376895844e-06,
+      "loss": 0.9512,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.395940343999691e-06,
+      "loss": 0.9273,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3918936048110197e-06,
+      "loss": 0.9301,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387849821695899e-06,
+      "loss": 1.0018,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3838089962255127e-06,
+      "loss": 0.9314,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3797711299698924e-06,
+      "loss": 0.9577,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375736224497922e-06,
+      "loss": 0.9036,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.371704281377335e-06,
+      "loss": 0.9547,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3676753021747113e-06,
+      "loss": 0.967,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.363649288455482e-06,
+      "loss": 0.91,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3596262417839256e-06,
+      "loss": 0.9322,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3556061637231653e-06,
+      "loss": 0.9475,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3515890558351773e-06,
+      "loss": 0.8916,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.347574919680773e-06,
+      "loss": 0.9303,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.343563756819619e-06,
+      "loss": 0.8944,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.9515,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3355503572099324e-06,
+      "loss": 0.9264,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3315481235749484e-06,
+      "loss": 0.8949,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3275488694603076e-06,
+      "loss": 0.8894,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323552596419889e-06,
+      "loss": 0.9382,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.319559306006417e-06,
+      "loss": 0.9222,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315568999771455e-06,
+      "loss": 0.9541,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3115816792654057e-06,
+      "loss": 0.9193,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3075973460375134e-06,
+      "loss": 0.9173,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.303616001635861e-06,
+      "loss": 0.907,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2996376476073724e-06,
+      "loss": 0.904,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.295662285497805e-06,
+      "loss": 0.958,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291689916851758e-06,
+      "loss": 0.901,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287720543212666e-06,
+      "loss": 0.8728,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2837541661228024e-06,
+      "loss": 0.9281,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2797907871232673e-06,
+      "loss": 0.9358,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.275830407754006e-06,
+      "loss": 0.8646,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.271873029553795e-06,
+      "loss": 0.9345,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267918654060244e-06,
+      "loss": 0.9303,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2639672828097956e-06,
+      "loss": 0.9069,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2600189173377263e-06,
+      "loss": 0.9434,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256073559178145e-06,
+      "loss": 0.9739,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2521312098639914e-06,
+      "loss": 0.9754,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2481918709270347e-06,
+      "loss": 0.9779,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2442555438978774e-06,
+      "loss": 0.8994,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.240322230305951e-06,
+      "loss": 0.8722,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2363919316795135e-06,
+      "loss": 0.9744,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2324646495456546e-06,
+      "loss": 0.9186,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2285403854302912e-06,
+      "loss": 0.9391,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224619140858166e-06,
+      "loss": 0.9907,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2207009173528528e-06,
+      "loss": 0.9245,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2167857164367424e-06,
+      "loss": 0.9369,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2128735396310606e-06,
+      "loss": 0.9688,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2089643884558553e-06,
+      "loss": 0.9263,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2050582644299976e-06,
+      "loss": 0.8944,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.201155169071184e-06,
+      "loss": 0.9504,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1972551038959313e-06,
+      "loss": 0.967,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1933580704195835e-06,
+      "loss": 0.9745,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189464070156302e-06,
+      "loss": 0.8504,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1855731046190732e-06,
+      "loss": 1.003,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.9283,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.177800283768817e-06,
+      "loss": 0.9378,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173918431475861e-06,
+      "loss": 0.9295,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1700396199491015e-06,
+      "loss": 0.9599,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1661638506956208e-06,
+      "loss": 0.926,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1622911252213195e-06,
+      "loss": 0.9643,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1584214450309194e-06,
+      "loss": 0.9188,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1545548116279557e-06,
+      "loss": 0.9202,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1506912265147772e-06,
+      "loss": 0.9881,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.146830691192553e-06,
+      "loss": 0.9773,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1429732071612653e-06,
+      "loss": 0.9214,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1391187759197118e-06,
+      "loss": 0.8604,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1352673989655026e-06,
+      "loss": 0.856,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.131419077795063e-06,
+      "loss": 0.9612,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1275738139036274e-06,
+      "loss": 0.9671,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1237316087852465e-06,
+      "loss": 0.8985,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119892463932781e-06,
+      "loss": 0.9268,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116056380837902e-06,
+      "loss": 0.9804,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1122233609910903e-06,
+      "loss": 0.8885,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.108393405881638e-06,
+      "loss": 0.8703,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.104566516997647e-06,
+      "loss": 0.8771,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1007426958260245e-06,
+      "loss": 0.9079,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.096921943852489e-06,
+      "loss": 0.9475,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093104262561569e-06,
+      "loss": 0.9487,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08928965343659e-06,
+      "loss": 0.9252,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0854781179596937e-06,
+      "loss": 0.9129,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0816696576118223e-06,
+      "loss": 0.9443,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077864273872727e-06,
+      "loss": 1.0086,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0740619682209607e-06,
+      "loss": 0.906,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0702627421338817e-06,
+      "loss": 0.9054,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0664665970876496e-06,
+      "loss": 0.9524,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06267353455723e-06,
+      "loss": 0.9681,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05888355601639e-06,
+      "loss": 0.9096,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0550966629376955e-06,
+      "loss": 0.9733,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0513128567925177e-06,
+      "loss": 0.9379,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0475321390510262e-06,
+      "loss": 0.8657,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043754511182191e-06,
+      "loss": 0.9505,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0399799746537806e-06,
+      "loss": 0.9211,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036208530932363e-06,
+      "loss": 0.9956,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0324401814833074e-06,
+      "loss": 0.9432,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0286749277707783e-06,
+      "loss": 0.9437,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.024912771257732e-06,
+      "loss": 0.9965,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021153713405931e-06,
+      "loss": 0.8993,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0173977556759284e-06,
+      "loss": 0.9559,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013644899527074e-06,
+      "loss": 0.9506,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.009895146417512e-06,
+      "loss": 0.971,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0061484978041813e-06,
+      "loss": 0.9808,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002404955142815e-06,
+      "loss": 0.9407,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9986645198879385e-06,
+      "loss": 0.9135,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9949271934928703e-06,
+      "loss": 1.0496,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9911929774097216e-06,
+      "loss": 1.0192,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9874618730893947e-06,
+      "loss": 0.9347,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983733881981581e-06,
+      "loss": 0.9279,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9800090055347654e-06,
+      "loss": 0.9493,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9762872451962214e-06,
+      "loss": 0.9198,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9725686024120093e-06,
+      "loss": 0.9935,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9688530786269854e-06,
+      "loss": 0.9519,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.965140675284779e-06,
+      "loss": 0.9361,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.961431393827827e-06,
+      "loss": 0.9455,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.95772523569734e-06,
+      "loss": 0.9514,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9540222023333165e-06,
+      "loss": 0.9614,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.950322295174545e-06,
+      "loss": 1.0095,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9466255156585965e-06,
+      "loss": 0.898,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9429318652218277e-06,
+      "loss": 0.9555,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9392413452993787e-06,
+      "loss": 0.9102,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9355539573251737e-06,
+      "loss": 0.956,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.93186970273192e-06,
+      "loss": 0.9506,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.928188582951108e-06,
+      "loss": 0.8936,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9245105994130086e-06,
+      "loss": 0.9054,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920835753546679e-06,
+      "loss": 0.9207,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9171640467799478e-06,
+      "loss": 0.9071,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913495480539432e-06,
+      "loss": 0.9399,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.9855,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.906167775337404e-06,
+      "loss": 0.914,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9025086392230163e-06,
+      "loss": 0.927,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8988526493290948e-06,
+      "loss": 0.9135,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.895199807076148e-06,
+      "loss": 0.9753,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.891550113883458e-06,
+      "loss": 0.9486,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8879035711690897e-06,
+      "loss": 0.9467,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8842601803498772e-06,
+      "loss": 0.9308,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.880619942841435e-06,
+      "loss": 0.9504,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8769828600581497e-06,
+      "loss": 0.9516,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8733489334131827e-06,
+      "loss": 0.899,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8697181643184703e-06,
+      "loss": 0.9468,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8660905541847208e-06,
+      "loss": 0.9834,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8624661044214154e-06,
+      "loss": 0.9165,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.858844816436809e-06,
+      "loss": 0.9112,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8552266916379225e-06,
+      "loss": 0.9044,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8516117314305526e-06,
+      "loss": 0.9331,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847999937219266e-06,
+      "loss": 0.9479,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8443913104073984e-06,
+      "loss": 0.9462,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8407858523970568e-06,
+      "loss": 0.9724,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8371835645891134e-06,
+      "loss": 0.8517,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.833584448383211e-06,
+      "loss": 0.8574,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82998850517776e-06,
+      "loss": 0.9482,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.826395736369937e-06,
+      "loss": 0.9957,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8228061433556866e-06,
+      "loss": 0.9436,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8192197275297174e-06,
+      "loss": 0.938,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.815636490285506e-06,
+      "loss": 0.9757,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8120564330152924e-06,
+      "loss": 0.8745,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.808479557110081e-06,
+      "loss": 0.8819,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8049058639596418e-06,
+      "loss": 0.9918,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801335354952507e-06,
+      "loss": 0.9385,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7977680314759693e-06,
+      "loss": 0.9492,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7942038949160857e-06,
+      "loss": 0.931,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7906429466576768e-06,
+      "loss": 0.9128,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7870851880843233e-06,
+      "loss": 0.9093,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7835306205783643e-06,
+      "loss": 0.9124,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7799792455209019e-06,
+      "loss": 0.9631,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776431064291796e-06,
+      "loss": 0.9399,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7728860782696666e-06,
+      "loss": 0.9185,
+      "step": 4235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7693442888318923e-06,
+      "loss": 0.9248,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.765805697354608e-06,
+      "loss": 0.9267,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7622703052127078e-06,
+      "loss": 0.9697,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7587381137798432e-06,
+      "loss": 0.9694,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7552091244284197e-06,
+      "loss": 0.9004,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7516833385296016e-06,
+      "loss": 0.8974,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7481607574533044e-06,
+      "loss": 0.9347,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.744641382568204e-06,
+      "loss": 0.9361,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741125215241728e-06,
+      "loss": 0.9444,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 0.9586,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7341025087281149e-06,
+      "loss": 0.9398,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.730595972269602e-06,
+      "loss": 0.9257,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7270926488269502e-06,
+      "loss": 0.9148,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7235925397613529e-06,
+      "loss": 0.9739,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7200956464327512e-06,
+      "loss": 0.9303,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.716601970199836e-06,
+      "loss": 0.9442,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7131115124200494e-06,
+      "loss": 0.9306,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709624274449584e-06,
+      "loss": 0.9454,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061402576433794e-06,
+      "loss": 0.9222,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026594633551252e-06,
+      "loss": 0.8613,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6991818929372584e-06,
+      "loss": 1.0201,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6957075477409623e-06,
+      "loss": 0.8554,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6922364291161697e-06,
+      "loss": 0.9796,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.688768538411556e-06,
+      "loss": 0.9011,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.9354,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6818424461513129e-06,
+      "loss": 0.9427,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6783842472867607e-06,
+      "loss": 0.9387,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.674929281724551e-06,
+      "loss": 0.9815,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6714775508070868e-06,
+      "loss": 0.9364,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6680290558755119e-06,
+      "loss": 0.8432,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6645837982697122e-06,
+      "loss": 0.9584,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6611417793283192e-06,
+      "loss": 0.9339,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6577030003887018e-06,
+      "loss": 0.9795,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6542674627869738e-06,
+      "loss": 0.9509,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6508351678579882e-06,
+      "loss": 1.0122,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6474061169353362e-06,
+      "loss": 0.9644,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6439803113513509e-06,
+      "loss": 0.8987,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640557752437103e-06,
+      "loss": 0.8913,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6371384415224046e-06,
+      "loss": 0.9972,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6337223799358025e-06,
+      "loss": 0.927,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6303095690045822e-06,
+      "loss": 0.9658,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6269000100547682e-06,
+      "loss": 0.9663,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6234937044111155e-06,
+      "loss": 0.91,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6200906533971206e-06,
+      "loss": 0.9164,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6166908583350138e-06,
+      "loss": 0.9339,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6132943205457607e-06,
+      "loss": 0.8433,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6099010413490624e-06,
+      "loss": 0.9552,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6065110220633495e-06,
+      "loss": 0.9857,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6031242640057908e-06,
+      "loss": 0.9255,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.599740768492286e-06,
+      "loss": 0.9744,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.596360536837468e-06,
+      "loss": 0.9368,
+      "step": 4286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5929835703546992e-06,
+      "loss": 0.8924,
+      "step": 4287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.589609870356076e-06,
+      "loss": 0.9091,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5862394381524239e-06,
+      "loss": 0.9328,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.582872275053301e-06,
+      "loss": 0.9557,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.579508382366992e-06,
+      "loss": 0.8853,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5761477614005127e-06,
+      "loss": 0.991,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5727904134596084e-06,
+      "loss": 0.9695,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5694363398487545e-06,
+      "loss": 0.9397,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.566085541871145e-06,
+      "loss": 0.8909,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5627380208287114e-06,
+      "loss": 0.9599,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5593937780221092e-06,
+      "loss": 0.9406,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556052814750717e-06,
+      "loss": 0.9917,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5527151323126422e-06,
+      "loss": 0.9501,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5493807320047183e-06,
+      "loss": 0.9859,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5460496151225002e-06,
+      "loss": 0.9493,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.542721782960268e-06,
+      "loss": 0.9736,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5393972368110287e-06,
+      "loss": 0.9352,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.536075977966509e-06,
+      "loss": 0.8904,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.8949,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5294433273521525e-06,
+      "loss": 0.9474,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5261319381593831e-06,
+      "loss": 0.9148,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.522823841425467e-06,
+      "loss": 0.9723,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5195190384357405e-06,
+      "loss": 0.8881,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5162175304742633e-06,
+      "loss": 0.9668,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5129193188238068e-06,
+      "loss": 0.8948,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5096244047658704e-06,
+      "loss": 0.9688,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5063327895806668e-06,
+      "loss": 0.9207,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5030444745471295e-06,
+      "loss": 0.95,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.499759460942909e-06,
+      "loss": 0.943,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4964777500443728e-06,
+      "loss": 0.8992,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4931993431266056e-06,
+      "loss": 0.9077,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4899242414634097e-06,
+      "loss": 0.905,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.486652446327299e-06,
+      "loss": 0.978,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4833839589895072e-06,
+      "loss": 0.9843,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.48011878071998e-06,
+      "loss": 0.9649,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.476856912787379e-06,
+      "loss": 0.9584,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4735983564590784e-06,
+      "loss": 0.9679,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.470343113001167e-06,
+      "loss": 0.9512,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.467091183678444e-06,
+      "loss": 0.9535,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4638425697544235e-06,
+      "loss": 0.9209,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.460597272491333e-06,
+      "loss": 0.9202,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4573552931501046e-06,
+      "loss": 0.9608,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4541166329903856e-06,
+      "loss": 0.9303,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4508812932705364e-06,
+      "loss": 0.9247,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4476492752476223e-06,
+      "loss": 0.9371,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4444205801774204e-06,
+      "loss": 0.8867,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4411952093144167e-06,
+      "loss": 0.952,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4379731639118045e-06,
+      "loss": 0.9001,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4347544452214869e-06,
+      "loss": 0.9566,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4315390544940732e-06,
+      "loss": 0.9391,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4283269929788779e-06,
+      "loss": 0.9879,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4251182619239257e-06,
+      "loss": 0.95,
+      "step": 4338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4219128625759426e-06,
+      "loss": 0.9827,
+      "step": 4339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4187107961803704e-06,
+      "loss": 0.9134,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4155120639813392e-06,
+      "loss": 0.9215,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4123166672216971e-06,
+      "loss": 0.9716,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.409124607142992e-06,
+      "loss": 0.891,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4059358849854732e-06,
+      "loss": 0.9188,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4027505019880972e-06,
+      "loss": 0.9639,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3995684593885206e-06,
+      "loss": 0.9331,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3963897584231023e-06,
+      "loss": 0.9172,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3932144003269033e-06,
+      "loss": 0.9292,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3900423863336842e-06,
+      "loss": 0.8987,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3868737176759105e-06,
+      "loss": 0.997,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3837083955847418e-06,
+      "loss": 0.9231,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3805464212900432e-06,
+      "loss": 0.9404,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.377387796020374e-06,
+      "loss": 0.941,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3742325210029973e-06,
+      "loss": 1.0015,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3710805974638697e-06,
+      "loss": 0.9225,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3679320266276518e-06,
+      "loss": 0.9074,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.364786809717692e-06,
+      "loss": 0.8721,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3616449479560434e-06,
+      "loss": 0.936,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3585064425634542e-06,
+      "loss": 0.9659,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3553712947593655e-06,
+      "loss": 0.8832,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3522395057619186e-06,
+      "loss": 0.9636,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3491110767879444e-06,
+      "loss": 1.0002,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.345986009052972e-06,
+      "loss": 0.9606,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342864303771223e-06,
+      "loss": 1.0479,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.9926,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3366309854177517e-06,
+      "loss": 0.9594,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3335193747679387e-06,
+      "loss": 0.9843,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.330411131415169e-06,
+      "loss": 0.9771,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3273062565671258e-06,
+      "loss": 0.9214,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.324204751430186e-06,
+      "loss": 0.9424,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3211066172094178e-06,
+      "loss": 0.9513,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3180118551085763e-06,
+      "loss": 0.9827,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3149204663301118e-06,
+      "loss": 0.9219,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3118324520751558e-06,
+      "loss": 0.9789,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3087478135435361e-06,
+      "loss": 0.9197,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056665519337663e-06,
+      "loss": 0.8665,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3025886684430467e-06,
+      "loss": 0.9384,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299514164267267e-06,
+      "loss": 0.9616,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2964430406010032e-06,
+      "loss": 0.9195,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.293375298637518e-06,
+      "loss": 0.9288,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2903109395687597e-06,
+      "loss": 0.8513,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2872499645853609e-06,
+      "loss": 0.9017,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2841923748766427e-06,
+      "loss": 0.9122,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2811381716306092e-06,
+      "loss": 0.9419,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.278087356033947e-06,
+      "loss": 0.9728,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2750399292720284e-06,
+      "loss": 0.9541,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271995892528909e-06,
+      "loss": 0.9402,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2689552469873267e-06,
+      "loss": 0.9391,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2659179938287035e-06,
+      "loss": 0.9844,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2628841342331389e-06,
+      "loss": 0.9835,
+      "step": 4390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2598536693794171e-06,
+      "loss": 0.9976,
+      "step": 4391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2568266004450037e-06,
+      "loss": 0.9416,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2538029286060428e-06,
+      "loss": 0.8641,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2507826550373614e-06,
+      "loss": 0.9389,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2477657809124632e-06,
+      "loss": 0.9298,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.244752307403534e-06,
+      "loss": 0.9257,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2417422356814345e-06,
+      "loss": 0.8998,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2387355669157076e-06,
+      "loss": 0.9384,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2357323022745716e-06,
+      "loss": 0.9229,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2327324429249232e-06,
+      "loss": 0.9431,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2297359900323346e-06,
+      "loss": 0.9117,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2267429447610557e-06,
+      "loss": 0.8905,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2237533082740127e-06,
+      "loss": 0.9508,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2207670817328065e-06,
+      "loss": 0.9011,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2177842662977136e-06,
+      "loss": 0.9596,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2148048631276876e-06,
+      "loss": 0.9414,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2118288733803474e-06,
+      "loss": 0.9067,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2088562982119957e-06,
+      "loss": 0.9362,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2058871387776039e-06,
+      "loss": 0.9263,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2029213962308172e-06,
+      "loss": 0.9694,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.199959071723953e-06,
+      "loss": 0.9175,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.197000166408001e-06,
+      "loss": 0.9694,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.19404468143262e-06,
+      "loss": 0.9804,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1910926179461446e-06,
+      "loss": 0.9103,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188143977095576e-06,
+      "loss": 0.9291,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185198760026587e-06,
+      "loss": 0.9604,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1822569678835195e-06,
+      "loss": 0.9613,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.179318601809386e-06,
+      "loss": 0.8988,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176383662945866e-06,
+      "loss": 0.9457,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1734521524333087e-06,
+      "loss": 1.0156,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1705240714107301e-06,
+      "loss": 0.8968,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.167599421015818e-06,
+      "loss": 0.9334,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.164678202384919e-06,
+      "loss": 0.9564,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1617604166530527e-06,
+      "loss": 0.981,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1588460649539036e-06,
+      "loss": 0.9295,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1559351484198211e-06,
+      "loss": 0.9351,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.153027668181821e-06,
+      "loss": 0.8712,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1501236253695823e-06,
+      "loss": 0.9707,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1472230211114498e-06,
+      "loss": 0.9559,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1443258565344329e-06,
+      "loss": 0.9324,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1414321327642019e-06,
+      "loss": 0.9445,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1385418509250933e-06,
+      "loss": 0.9029,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1356550121401033e-06,
+      "loss": 1.0011,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1327716175308934e-06,
+      "loss": 0.8934,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.129891668217783e-06,
+      "loss": 1.0232,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1270151653197569e-06,
+      "loss": 0.9376,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124142109954459e-06,
+      "loss": 0.9265,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1212725032381933e-06,
+      "loss": 0.8845,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1184063462859262e-06,
+      "loss": 0.9103,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1155436402112785e-06,
+      "loss": 0.9675,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1126843861265347e-06,
+      "loss": 0.9439,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1098285851426372e-06,
+      "loss": 0.9282,
+      "step": 4442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1069762383691862e-06,
+      "loss": 0.8968,
+      "step": 4443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1041273469144399e-06,
+      "loss": 0.9444,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1012819118853147e-06,
+      "loss": 0.9429,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.098439934387383e-06,
+      "loss": 0.9791,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.095601415524874e-06,
+      "loss": 0.9147,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.092766356400673e-06,
+      "loss": 0.9664,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 0.9304,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0871066217720173e-06,
+      "loss": 0.9379,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.08428194846661e-06,
+      "loss": 0.9568,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0814607392976062e-06,
+      "loss": 0.9142,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0786429953611665e-06,
+      "loss": 0.9851,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0758287177521032e-06,
+      "loss": 0.9392,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073017907563887e-06,
+      "loss": 0.9128,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0702105658886318e-06,
+      "loss": 0.9725,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0674066938171123e-06,
+      "loss": 0.9484,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0646062924387513e-06,
+      "loss": 0.8408,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061809362841626e-06,
+      "loss": 0.9239,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0590159061124606e-06,
+      "loss": 0.9003,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0562259233366334e-06,
+      "loss": 0.8766,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.053439415598171e-06,
+      "loss": 0.8836,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0506563839797501e-06,
+      "loss": 0.9105,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0478768295626985e-06,
+      "loss": 0.9714,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0451007534269908e-06,
+      "loss": 0.9124,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.04232815665125e-06,
+      "loss": 0.9081,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0395590403127487e-06,
+      "loss": 0.9908,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0367934054874074e-06,
+      "loss": 0.8905,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.034031253249792e-06,
+      "loss": 0.9683,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0312725846731174e-06,
+      "loss": 0.9393,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0285174008292431e-06,
+      "loss": 0.9618,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.025765702788677e-06,
+      "loss": 0.8732,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0230174916205681e-06,
+      "loss": 0.9061,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0202727683927149e-06,
+      "loss": 0.9721,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0175315341715598e-06,
+      "loss": 0.813,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0147937900221882e-06,
+      "loss": 0.918,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.012059537008332e-06,
+      "loss": 0.8589,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.009328776192363e-06,
+      "loss": 0.9626,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0066015086353e-06,
+      "loss": 0.9265,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.003877735396801e-06,
+      "loss": 0.8762,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0011574575351702e-06,
+      "loss": 0.9458,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.984406761073496e-07,
+      "loss": 1.0192,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.957273921689237e-07,
+      "loss": 0.8951,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.930176067741216e-07,
+      "loss": 0.8851,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.9175,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.87608535825494e-07,
+      "loss": 0.9453,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.849092523733217e-07,
+      "loss": 0.9581,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.822134716680832e-07,
+      "loss": 0.9301,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.795211947571993e-07,
+      "loss": 0.9796,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.768324226867353e-07,
+      "loss": 0.9598,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74147156501396e-07,
+      "loss": 0.9094,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.714653972445199e-07,
+      "loss": 0.999,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.687871459580845e-07,
+      "loss": 0.9278,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.661124036827064e-07,
+      "loss": 0.9056,
+      "step": 4494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.634411714576353e-07,
+      "loss": 0.9301,
+      "step": 4495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.60773450320759e-07,
+      "loss": 0.8941,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.58109241308599e-07,
+      "loss": 1.0095,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.55448545456319e-07,
+      "loss": 0.9222,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.527913637977093e-07,
+      "loss": 0.9659,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.501376973651999e-07,
+      "loss": 0.8875,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.474875471898526e-07,
+      "loss": 0.9478,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.448409143013626e-07,
+      "loss": 0.8809,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.421977997280596e-07,
+      "loss": 0.8732,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.395582044969076e-07,
+      "loss": 0.9584,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.369221296335007e-07,
+      "loss": 0.8582,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.342895761620663e-07,
+      "loss": 0.9229,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.31660545105465e-07,
+      "loss": 0.9414,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.290350374851865e-07,
+      "loss": 0.8885,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.264130543213512e-07,
+      "loss": 0.9438,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.237945966327133e-07,
+      "loss": 0.9768,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.211796654366545e-07,
+      "loss": 0.9191,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.185682617491865e-07,
+      "loss": 0.9038,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.15960386584952e-07,
+      "loss": 0.9477,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.133560409572195e-07,
+      "loss": 0.9724,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.107552258778907e-07,
+      "loss": 0.9956,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.081579423574916e-07,
+      "loss": 0.9178,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.055641914051783e-07,
+      "loss": 0.8211,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.029739740287358e-07,
+      "loss": 0.9891,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.00387291234569e-07,
+      "loss": 0.955,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.978041440277163e-07,
+      "loss": 0.8588,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.952245334118415e-07,
+      "loss": 0.9851,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.926484603892327e-07,
+      "loss": 0.9401,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.900759259608038e-07,
+      "loss": 0.9903,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.875069311260942e-07,
+      "loss": 0.9338,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.849414768832687e-07,
+      "loss": 0.9732,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.823795642291145e-07,
+      "loss": 0.9371,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79821194159044e-07,
+      "loss": 0.8959,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.772663676670934e-07,
+      "loss": 0.9207,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.747150857459219e-07,
+      "loss": 0.9888,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.721673493868111e-07,
+      "loss": 0.9554,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.696231595796633e-07,
+      "loss": 0.9727,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.670825173130081e-07,
+      "loss": 0.9477,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.909,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.620118793483823e-07,
+      "loss": 0.959,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.5948188562057e-07,
+      "loss": 0.9408,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.56955443373565e-07,
+      "loss": 0.954,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.544325535889997e-07,
+      "loss": 0.9759,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.519132172471212e-07,
+      "loss": 0.8737,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.493974353268019e-07,
+      "loss": 0.9111,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.468852088055291e-07,
+      "loss": 0.9161,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.443765386594094e-07,
+      "loss": 0.9582,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.418714258631689e-07,
+      "loss": 0.9202,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.393698713901499e-07,
+      "loss": 0.933,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.36871876212314e-07,
+      "loss": 0.8755,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.343774413002382e-07,
+      "loss": 0.8656,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.318865676231158e-07,
+      "loss": 0.8984,
+      "step": 4546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.293992561487596e-07,
+      "loss": 0.9805,
+      "step": 4547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.26915507843593e-07,
+      "loss": 0.9779,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.24435323672661e-07,
+      "loss": 0.9005,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.219587045996203e-07,
+      "loss": 0.9876,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.194856515867433e-07,
+      "loss": 0.9766,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.170161655949139e-07,
+      "loss": 0.8626,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.145502475836331e-07,
+      "loss": 0.883,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.120878985110181e-07,
+      "loss": 1.0051,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.096291193337935e-07,
+      "loss": 0.8781,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.071739110073018e-07,
+      "loss": 0.9497,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.047222744854943e-07,
+      "loss": 0.9264,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.022742107209369e-07,
+      "loss": 0.9421,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.998297206648075e-07,
+      "loss": 0.8942,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.973888052668943e-07,
+      "loss": 0.9547,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.949514654755963e-07,
+      "loss": 0.896,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.925177022379249e-07,
+      "loss": 0.8957,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.900875164995003e-07,
+      "loss": 0.9732,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.876609092045528e-07,
+      "loss": 0.9619,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.852378812959227e-07,
+      "loss": 0.916,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.828184337150613e-07,
+      "loss": 0.9748,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.804025674020266e-07,
+      "loss": 0.9419,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.779902832954833e-07,
+      "loss": 0.9089,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.755815823327084e-07,
+      "loss": 0.9478,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.731764654495832e-07,
+      "loss": 0.9716,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.707749335806014e-07,
+      "loss": 0.9272,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.683769876588576e-07,
+      "loss": 0.8731,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.659826286160565e-07,
+      "loss": 0.9037,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.635918573825096e-07,
+      "loss": 0.8663,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.8982,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.588210820574482e-07,
+      "loss": 0.8636,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.564410798195832e-07,
+      "loss": 0.9514,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.540646690982711e-07,
+      "loss": 0.8634,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.516918508168469e-07,
+      "loss": 0.8763,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.493226258972519e-07,
+      "loss": 0.9442,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46956995260033e-07,
+      "loss": 0.9572,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.445949598243362e-07,
+      "loss": 0.9187,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.422365205079141e-07,
+      "loss": 0.9635,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.398816782271223e-07,
+      "loss": 0.9471,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.375304338969135e-07,
+      "loss": 0.9767,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.351827884308493e-07,
+      "loss": 0.9012,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.328387427410877e-07,
+      "loss": 0.9745,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.304982977383935e-07,
+      "loss": 0.9696,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281614543321269e-07,
+      "loss": 0.9206,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.258282134302519e-07,
+      "loss": 0.8849,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.234985759393321e-07,
+      "loss": 0.8966,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.211725427645311e-07,
+      "loss": 0.9054,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.188501148096117e-07,
+      "loss": 0.9657,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.165312929769352e-07,
+      "loss": 0.9093,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.142160781674645e-07,
+      "loss": 0.9002,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.119044712807577e-07,
+      "loss": 0.9222,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.095964732149741e-07,
+      "loss": 0.9417,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.072920848668674e-07,
+      "loss": 0.9639,
+      "step": 4598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.049913071317927e-07,
+      "loss": 0.9313,
+      "step": 4599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.026941409036991e-07,
+      "loss": 0.9298,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.004005870751341e-07,
+      "loss": 0.9438,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.981106465372389e-07,
+      "loss": 0.9002,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.958243201797554e-07,
+      "loss": 0.8976,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.935416088910185e-07,
+      "loss": 0.9271,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.912625135579587e-07,
+      "loss": 0.9106,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.889870350661009e-07,
+      "loss": 0.9592,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.867151742995659e-07,
+      "loss": 0.8353,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.844469321410685e-07,
+      "loss": 0.978,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.821823094719171e-07,
+      "loss": 0.8286,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.799213071720156e-07,
+      "loss": 0.9676,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.776639261198581e-07,
+      "loss": 0.893,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.754101671925339e-07,
+      "loss": 1.0181,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.731600312657238e-07,
+      "loss": 0.8608,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.709135192137018e-07,
+      "loss": 0.9095,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.68670631909335e-07,
+      "loss": 0.9527,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.664313702240788e-07,
+      "loss": 0.896,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.641957350279838e-07,
+      "loss": 0.8888,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.619637271896873e-07,
+      "loss": 0.9072,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.597353475764201e-07,
+      "loss": 0.9498,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.57510597054003e-07,
+      "loss": 0.9174,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.552894764868456e-07,
+      "loss": 0.9361,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.530719867379498e-07,
+      "loss": 0.9288,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.508581286689031e-07,
+      "loss": 0.8687,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.48647903139884e-07,
+      "loss": 0.9475,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.464413110096601e-07,
+      "loss": 0.9542,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.442383531355856e-07,
+      "loss": 1.0222,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.420390303736034e-07,
+      "loss": 0.9438,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.398433435782459e-07,
+      "loss": 0.8699,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.37651293602628e-07,
+      "loss": 0.9002,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.354628812984576e-07,
+      "loss": 0.9135,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.332781075160244e-07,
+      "loss": 0.841,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.31096973104206e-07,
+      "loss": 0.934,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.28919478910468e-07,
+      "loss": 0.9393,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.26745625780858e-07,
+      "loss": 0.9435,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.245754145600091e-07,
+      "loss": 0.9288,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.224088460911437e-07,
+      "loss": 0.9811,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.202459212160638e-07,
+      "loss": 0.9569,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.180866407751595e-07,
+      "loss": 0.917,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.15931005607403e-07,
+      "loss": 0.8238,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.137790165503499e-07,
+      "loss": 0.9592,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.116306744401391e-07,
+      "loss": 0.9746,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.09485980111495e-07,
+      "loss": 0.8856,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.073449343977211e-07,
+      "loss": 0.9744,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.052075381307054e-07,
+      "loss": 0.9707,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.9667,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.009436972574079e-07,
+      "loss": 0.936,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.988172543078097e-07,
+      "loss": 0.9281,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.966944641183369e-07,
+      "loss": 0.8984,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.945753275137844e-07,
+      "loss": 0.9405,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.924598453175278e-07,
+      "loss": 0.8712,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.903480183515197e-07,
+      "loss": 0.9395,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.882398474362949e-07,
+      "loss": 0.9209,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.861353333909692e-07,
+      "loss": 0.9615,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.840344770332351e-07,
+      "loss": 0.9396,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.819372791793654e-07,
+      "loss": 0.9395,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.798437406442104e-07,
+      "loss": 0.9408,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.777538622412005e-07,
+      "loss": 0.8503,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.756676447823406e-07,
+      "loss": 0.9522,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.735850890782158e-07,
+      "loss": 0.9328,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.715061959379875e-07,
+      "loss": 0.8291,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.694309661693942e-07,
+      "loss": 0.8911,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.67359400578752e-07,
+      "loss": 0.9165,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.652914999709535e-07,
+      "loss": 0.9213,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.632272651494631e-07,
+      "loss": 0.8934,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.8355,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.591097960721581e-07,
+      "loss": 0.8546,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.570565634161562e-07,
+      "loss": 0.9876,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.550069997460872e-07,
+      "loss": 0.9434,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.529611058582951e-07,
+      "loss": 0.9278,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509188825476964e-07,
+      "loss": 0.9183,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.488803306077806e-07,
+      "loss": 0.9334,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.468454508306131e-07,
+      "loss": 0.8862,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.448142440068316e-07,
+      "loss": 0.9669,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.427867109256457e-07,
+      "loss": 0.9251,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.407628523748398e-07,
+      "loss": 0.9471,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.387426691407682e-07,
+      "loss": 0.878,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367261620083575e-07,
+      "loss": 0.9134,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.347133317611075e-07,
+      "loss": 0.9192,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.327041791810905e-07,
+      "loss": 0.9059,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.306987050489442e-07,
+      "loss": 1.0042,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286969101438821e-07,
+      "loss": 0.9634,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266987952436875e-07,
+      "loss": 0.9782,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.247043611247127e-07,
+      "loss": 0.8899,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.227136085618801e-07,
+      "loss": 0.927,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.207265383286831e-07,
+      "loss": 0.8754,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.187431511971819e-07,
+      "loss": 0.9042,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.167634479380068e-07,
+      "loss": 0.8915,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.147874293203581e-07,
+      "loss": 0.9622,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.128150961120026e-07,
+      "loss": 0.9647,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.108464490792753e-07,
+      "loss": 0.9014,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.088814889870808e-07,
+      "loss": 0.8924,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.069202165988874e-07,
+      "loss": 0.8675,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.049626326767366e-07,
+      "loss": 0.9199,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.030087379812299e-07,
+      "loss": 0.917,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.010585332715401e-07,
+      "loss": 0.9537,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.991120193054056e-07,
+      "loss": 0.9222,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.97169196839129e-07,
+      "loss": 0.9554,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.952300666275789e-07,
+      "loss": 0.9679,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.932946294241903e-07,
+      "loss": 0.9131,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.913628859809638e-07,
+      "loss": 0.8868,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.9513,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875104833758204e-07,
+      "loss": 0.9079,
+      "step": 4702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.855898257107261e-07,
+      "loss": 0.9009,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.836728647994371e-07,
+      "loss": 0.9639,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.817596013867765e-07,
+      "loss": 0.91,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.798500362161273e-07,
+      "loss": 0.9133,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.779441700294362e-07,
+      "loss": 0.8917,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7604200356721644e-07,
+      "loss": 0.8551,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7414353756853773e-07,
+      "loss": 1.0063,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7224877277103673e-07,
+      "loss": 0.8997,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703577099109091e-07,
+      "loss": 0.9293,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6847034972291595e-07,
+      "loss": 0.9331,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6658669294037393e-07,
+      "loss": 0.913,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.647067402951655e-07,
+      "loss": 0.9574,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.628304925177318e-07,
+      "loss": 0.9523,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.609579503370754e-07,
+      "loss": 0.9303,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5908911448075746e-07,
+      "loss": 0.9353,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.572239856749028e-07,
+      "loss": 0.9385,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553625646441928e-07,
+      "loss": 0.8906,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.535048521118668e-07,
+      "loss": 0.8968,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5165084879972844e-07,
+      "loss": 0.9412,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.498005554281337e-07,
+      "loss": 0.8881,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4795397271600314e-07,
+      "loss": 0.91,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.461111013808117e-07,
+      "loss": 0.8946,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 0.9231,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.424364957039384e-07,
+      "loss": 0.9324,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4060476278999897e-07,
+      "loss": 0.8562,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3877674410848005e-07,
+      "loss": 0.882,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3695244036964567e-07,
+      "loss": 0.9799,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.351318522823134e-07,
+      "loss": 0.9725,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3331498055386057e-07,
+      "loss": 0.9178,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3150182589021926e-07,
+      "loss": 0.9276,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.296923889958771e-07,
+      "loss": 0.9019,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.278866705738782e-07,
+      "loss": 0.915,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2608467132581934e-07,
+      "loss": 0.9778,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2428639195185585e-07,
+      "loss": 0.9438,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.224918331506955e-07,
+      "loss": 0.9623,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2070099561960064e-07,
+      "loss": 0.9573,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.189138800543868e-07,
+      "loss": 0.9094,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.171304871494264e-07,
+      "loss": 0.9405,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1535081759764286e-07,
+      "loss": 0.9127,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1357487209051416e-07,
+      "loss": 0.9309,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.118026513180695e-07,
+      "loss": 1.0002,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1003415596889363e-07,
+      "loss": 0.9353,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.082693867301224e-07,
+      "loss": 0.9634,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.065083442874418e-07,
+      "loss": 0.971,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0475102932509336e-07,
+      "loss": 0.9024,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029974425258687e-07,
+      "loss": 0.9122,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.012475845711106e-07,
+      "loss": 0.9251,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.99501456140714e-07,
+      "loss": 0.8531,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9775905791312407e-07,
+      "loss": 1.0223,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.96020390565337e-07,
+      "loss": 0.8092,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9428545477289913e-07,
+      "loss": 0.9688,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9255425120990564e-07,
+      "loss": 0.9168,
+      "step": 4754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.908267805490051e-07,
+      "loss": 0.9832,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8910304346139185e-07,
+      "loss": 0.9426,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8738304061681107e-07,
+      "loss": 0.9349,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8566677268355943e-07,
+      "loss": 0.9692,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8395424032847686e-07,
+      "loss": 0.9803,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8224544421695766e-07,
+      "loss": 0.8701,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8054038501294077e-07,
+      "loss": 0.9384,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.788390633789163e-07,
+      "loss": 0.966,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7714147997591655e-07,
+      "loss": 0.9347,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7544763546352834e-07,
+      "loss": 0.9773,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7375753049987974e-07,
+      "loss": 0.9272,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7207116574165094e-07,
+      "loss": 0.9514,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.703885418440656e-07,
+      "loss": 0.9531,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.687096594608941e-07,
+      "loss": 0.9342,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6703451924445467e-07,
+      "loss": 0.8886,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6536312184560996e-07,
+      "loss": 0.9384,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6369546791377054e-07,
+      "loss": 0.9681,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.620315580968892e-07,
+      "loss": 0.9374,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.603713930414676e-07,
+      "loss": 0.9124,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.587149733925488e-07,
+      "loss": 0.9158,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.570622997937234e-07,
+      "loss": 0.9418,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5541337288712564e-07,
+      "loss": 0.8889,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5376819331343404e-07,
+      "loss": 0.9031,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5212676171187065e-07,
+      "loss": 0.97,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.50489078720202e-07,
+      "loss": 0.9141,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4885514497473574e-07,
+      "loss": 0.9296,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472249611103273e-07,
+      "loss": 0.9029,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.455985277603713e-07,
+      "loss": 0.9678,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4397584555680653e-07,
+      "loss": 0.9871,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4235691513011425e-07,
+      "loss": 0.8458,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.9525,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3913031212198356e-07,
+      "loss": 0.8924,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.375226407942178e-07,
+      "loss": 0.8901,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.359187237506689e-07,
+      "loss": 0.9312,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3431856161452835e-07,
+      "loss": 0.9267,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.327221550075266e-07,
+      "loss": 0.9173,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3112950454993633e-07,
+      "loss": 0.9812,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2954061086056924e-07,
+      "loss": 0.9062,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2795547455677813e-07,
+      "loss": 0.9993,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.26374096254457e-07,
+      "loss": 0.8897,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.247964765680389e-07,
+      "loss": 0.9506,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2322261611049457e-07,
+      "loss": 0.9005,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2165251549333585e-07,
+      "loss": 0.9029,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.20086175326616e-07,
+      "loss": 0.9063,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.185235962189237e-07,
+      "loss": 1.0023,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1696477877738664e-07,
+      "loss": 0.9183,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1540972360767254e-07,
+      "loss": 0.8823,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1385843131398697e-07,
+      "loss": 0.9737,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1231090249907336e-07,
+      "loss": 0.8544,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.107671377642119e-07,
+      "loss": 0.9699,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0922713770922155e-07,
+      "loss": 0.9686,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.076909029324571e-07,
+      "loss": 0.9633,
+      "step": 4806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.06158434030811e-07,
+      "loss": 0.9848,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0462973159971375e-07,
+      "loss": 0.9431,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0310479623313125e-07,
+      "loss": 0.9473,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0158362852356627e-07,
+      "loss": 0.9401,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0006622906205707e-07,
+      "loss": 0.9225,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.985525984381754e-07,
+      "loss": 0.9488,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.970427372400353e-07,
+      "loss": 0.9718,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.955366460542797e-07,
+      "loss": 0.9135,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.940343254660905e-07,
+      "loss": 0.9512,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.925357760591818e-07,
+      "loss": 0.9165,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.910409984158058e-07,
+      "loss": 0.973,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8954999311674557e-07,
+      "loss": 0.8967,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.880627607413211e-07,
+      "loss": 0.9676,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.865793018673857e-07,
+      "loss": 0.9093,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8509961707132496e-07,
+      "loss": 0.916,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8362370692806117e-07,
+      "loss": 0.9953,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.821515720110479e-07,
+      "loss": 0.9739,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.806832128922732e-07,
+      "loss": 0.8716,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7921863014225504e-07,
+      "loss": 0.8931,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7775782433004714e-07,
+      "loss": 0.9453,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7630079602323447e-07,
+      "loss": 0.9292,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7484754578793513e-07,
+      "loss": 0.9623,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733980741887987e-07,
+      "loss": 0.8704,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7195238178900685e-07,
+      "loss": 0.9811,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7051046915027244e-07,
+      "loss": 0.8845,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.690723368328407e-07,
+      "loss": 0.953,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.676379853954858e-07,
+      "loss": 0.9581,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.662074153955152e-07,
+      "loss": 0.9615,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6478062738876654e-07,
+      "loss": 0.8565,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6335762192960743e-07,
+      "loss": 0.9909,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6193839957093683e-07,
+      "loss": 0.9835,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.605229608641824e-07,
+      "loss": 0.9097,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5911130635930203e-07,
+      "loss": 0.9459,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.577034366047848e-07,
+      "loss": 0.8648,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5629935214764866e-07,
+      "loss": 1.0083,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5489905353343725e-07,
+      "loss": 0.8152,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5350254130622866e-07,
+      "loss": 0.8923,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5210981600862663e-07,
+      "loss": 0.9099,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.8962,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4933572836530307e-07,
+      "loss": 0.9367,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4795436709743384e-07,
+      "loss": 0.8715,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.465767949148734e-07,
+      "loss": 0.8843,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.45203012352867e-07,
+      "loss": 0.9615,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4383301994518773e-07,
+      "loss": 0.9054,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.424668182241363e-07,
+      "loss": 1.0088,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4110440772054155e-07,
+      "loss": 0.9535,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3974578896375555e-07,
+      "loss": 0.9123,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.383909624816616e-07,
+      "loss": 0.923,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.370399288006664e-07,
+      "loss": 0.9541,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3569268844570447e-07,
+      "loss": 0.941,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3434924194023712e-07,
+      "loss": 0.9182,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3300958980624787e-07,
+      "loss": 0.9554,
+      "step": 4858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3167373256425152e-07,
+      "loss": 0.9206,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3034167073328283e-07,
+      "loss": 0.9724,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2901340483090785e-07,
+      "loss": 0.8724,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2768893537321145e-07,
+      "loss": 0.9426,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2636826287480872e-07,
+      "loss": 0.9061,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2505138784883585e-07,
+      "loss": 0.872,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2373831080695463e-07,
+      "loss": 0.9377,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2242903225935254e-07,
+      "loss": 1.0121,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.211235527147404e-07,
+      "loss": 0.9679,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1982187268035138e-07,
+      "loss": 1.0023,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.874,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1722991316380005e-07,
+      "loss": 0.9399,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1593963468872435e-07,
+      "loss": 0.9538,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1465315773804617e-07,
+      "loss": 0.8994,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1337048281161565e-07,
+      "loss": 0.893,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1209161040780746e-07,
+      "loss": 0.9237,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1081654102351634e-07,
+      "loss": 0.958,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0954527515416156e-07,
+      "loss": 0.8945,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.082778132936858e-07,
+      "loss": 0.9595,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0701415593455065e-07,
+      "loss": 0.9519,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.057543035677412e-07,
+      "loss": 0.9796,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0449825668276246e-07,
+      "loss": 0.9609,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0324601576764525e-07,
+      "loss": 0.8864,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0199758130893478e-07,
+      "loss": 0.9698,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0075295379170413e-07,
+      "loss": 0.9245,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9951213369954204e-07,
+      "loss": 0.9384,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9827512151456175e-07,
+      "loss": 0.963,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9704191771739322e-07,
+      "loss": 0.9354,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.95812522787191e-07,
+      "loss": 0.8636,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9458693720162513e-07,
+      "loss": 0.9682,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.933651614368892e-07,
+      "loss": 0.8938,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.9041,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9093304126727497e-07,
+      "loss": 0.9615,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.897226978073774e-07,
+      "loss": 0.9446,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.885161660582746e-07,
+      "loss": 0.9426,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731344648875382e-07,
+      "loss": 0.8903,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8611453956612346e-07,
+      "loss": 0.8217,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8491944575621091e-07,
+      "loss": 0.9573,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8372816552336025e-07,
+      "loss": 0.9108,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8254069933043572e-07,
+      "loss": 0.9748,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.81357047638816e-07,
+      "loss": 0.896,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8017721090840324e-07,
+      "loss": 0.8746,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7900118959761181e-07,
+      "loss": 0.9409,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7782898416337735e-07,
+      "loss": 0.9203,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7666059506115108e-07,
+      "loss": 0.9603,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.754960227449032e-07,
+      "loss": 0.9747,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7433526766711727e-07,
+      "loss": 0.8789,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7317833027879928e-07,
+      "loss": 0.9754,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7202521102946624e-07,
+      "loss": 0.8811,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7087591036715534e-07,
+      "loss": 0.9393,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6973042873841827e-07,
+      "loss": 0.8934,
+      "step": 4909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6858876658832235e-07,
+      "loss": 0.8741,
+      "step": 4910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6745092436045495e-07,
+      "loss": 0.9631,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6631690249691358e-07,
+      "loss": 0.9533,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.651867014383146e-07,
+      "loss": 0.8907,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6406032162379127e-07,
+      "loss": 0.9353,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.629377634909868e-07,
+      "loss": 0.8659,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.618190274760656e-07,
+      "loss": 0.9301,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6070411401370335e-07,
+      "loss": 0.862,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5959302353709128e-07,
+      "loss": 0.9096,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.584857564779363e-07,
+      "loss": 0.9503,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5738231326645758e-07,
+      "loss": 0.8621,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5628269433139e-07,
+      "loss": 0.9017,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5518690009998282e-07,
+      "loss": 0.959,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5409493099799888e-07,
+      "loss": 0.9633,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.530067874497143e-07,
+      "loss": 0.9704,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.9612,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5084197870391836e-07,
+      "loss": 0.9965,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.497653143475275e-07,
+      "loss": 0.9386,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.486924772270759e-07,
+      "loss": 0.935,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4762346775940794e-07,
+      "loss": 0.8506,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.465582863598791e-07,
+      "loss": 0.9702,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4549693344235726e-07,
+      "loss": 0.9613,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.444394094192225e-07,
+      "loss": 0.9303,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4338571470137063e-07,
+      "loss": 0.9172,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4233584969820414e-07,
+      "loss": 0.9173,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4128981481764115e-07,
+      "loss": 1.0094,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4024761046611213e-07,
+      "loss": 0.879,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3920923704855648e-07,
+      "loss": 0.8769,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3817469496842596e-07,
+      "loss": 0.956,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3714398462768563e-07,
+      "loss": 0.9401,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3611710642681076e-07,
+      "loss": 0.923,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.350940607647866e-07,
+      "loss": 0.9077,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3407484803910854e-07,
+      "loss": 0.936,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3305946864578645e-07,
+      "loss": 0.9469,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.320479229793359e-07,
+      "loss": 0.8967,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3104021143278911e-07,
+      "loss": 0.9346,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3003633439768182e-07,
+      "loss": 0.9147,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2903629226406423e-07,
+      "loss": 0.89,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.280400854204944e-07,
+      "loss": 0.9231,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2704771425404382e-07,
+      "loss": 0.9326,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2605917915028743e-07,
+      "loss": 0.9607,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2507448049331571e-07,
+      "loss": 0.9684,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.24093618665726e-07,
+      "loss": 0.9363,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 0.9734,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.221434070216243e-07,
+      "loss": 0.9282,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2117405796285286e-07,
+      "loss": 0.9447,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2020854724894448e-07,
+      "loss": 0.939,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.192468752550402e-07,
+      "loss": 0.9223,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1828904235478888e-07,
+      "loss": 0.9743,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1733504892035064e-07,
+      "loss": 0.8579,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1638489532239339e-07,
+      "loss": 0.8652,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1543858193009183e-07,
+      "loss": 0.9282,
+      "step": 4961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1449610911112741e-07,
+      "loss": 0.8787,
+      "step": 4962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1355747723169386e-07,
+      "loss": 0.9279,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1262268665648724e-07,
+      "loss": 0.9631,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 0.9197,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1076463087008938e-07,
+      "loss": 0.9107,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0984136638083176e-07,
+      "loss": 0.8593,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0892194463966943e-07,
+      "loss": 1.0226,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0800636600383662e-07,
+      "loss": 0.8993,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0709463082907545e-07,
+      "loss": 0.9507,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0618673946963365e-07,
+      "loss": 0.9473,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.052826922782646e-07,
+      "loss": 0.9709,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0438248960623065e-07,
+      "loss": 0.8967,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0348613180329758e-07,
+      "loss": 0.9834,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0259361921774014e-07,
+      "loss": 0.9401,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0170495219633647e-07,
+      "loss": 0.9957,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0082013108437038e-07,
+      "loss": 0.9661,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.993915622563578e-08,
+      "loss": 0.9715,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.993915622563578e-08,
+      "loss": 0.8652,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.906202796242547e-08,
+      "loss": 0.8483,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.818874663554356e-08,
+      "loss": 1.0096,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.731931258429638e-08,
+      "loss": 0.9489,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.645372614649595e-08,
+      "loss": 0.9509,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.55919876584599e-08,
+      "loss": 0.9089,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.473409745500928e-08,
+      "loss": 1.0015,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.388005586947191e-08,
+      "loss": 0.9178,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.302986323368012e-08,
+      "loss": 0.9842,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.218351987796859e-08,
+      "loss": 0.8846,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.134102613117757e-08,
+      "loss": 0.9674,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.0502382320653e-08,
+      "loss": 0.9413,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.966758877224202e-08,
+      "loss": 0.9557,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.88366458102996e-08,
+      "loss": 0.9416,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.800955375768194e-08,
+      "loss": 0.9146,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.718631293574753e-08,
+      "loss": 0.9216,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.636692366436383e-08,
+      "loss": 0.9439,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.555138626189619e-08,
+      "loss": 0.9707,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.473970104521556e-08,
+      "loss": 0.8867,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.393186832969746e-08,
+      "loss": 0.9135,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.31278884292186e-08,
+      "loss": 0.9083,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.232776165616019e-08,
+      "loss": 0.9592,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.153148832140467e-08,
+      "loss": 0.9376,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.07390687343379e-08,
+      "loss": 0.89,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.995050320285025e-08,
+      "loss": 0.9539,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.916579203333107e-08,
+      "loss": 0.9021,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.838493553067539e-08,
+      "loss": 0.9079,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.760793399827937e-08,
+      "loss": 0.9863,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.683478773804043e-08,
+      "loss": 0.8668,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.606549705035937e-08,
+      "loss": 0.9542,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.53000622341371e-08,
+      "loss": 1.0081,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.453848358678018e-08,
+      "loss": 0.9509,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.378076140419188e-08,
+      "loss": 0.9642,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.302689598078117e-08,
+      "loss": 0.9675,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.227688760945595e-08,
+      "loss": 0.8845,
+      "step": 5013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.153073658162646e-08,
+      "loss": 0.8982,
+      "step": 5014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.078844318720523e-08,
+      "loss": 0.9007,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.00500077146038e-08,
+      "loss": 1.0095,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.931543045073708e-08,
+      "loss": 0.9063,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.858471168101788e-08,
+      "loss": 0.9179,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.785785168936244e-08,
+      "loss": 0.9459,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.713485075818815e-08,
+      "loss": 0.8951,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.641570916840923e-08,
+      "loss": 0.8974,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.570042719944436e-08,
+      "loss": 0.8958,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.498900512921013e-08,
+      "loss": 0.9023,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.428144323412544e-08,
+      "loss": 0.9287,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.357774178910703e-08,
+      "loss": 0.9025,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.287790106757396e-08,
+      "loss": 0.9187,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.21819213414443e-08,
+      "loss": 0.9706,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.148980288113504e-08,
+      "loss": 0.9394,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.080154595556331e-08,
+      "loss": 0.8902,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.011715083214742e-08,
+      "loss": 0.9479,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.943661777680354e-08,
+      "loss": 0.9512,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.875994705394794e-08,
+      "loss": 0.9305,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.808713892649587e-08,
+      "loss": 0.9207,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7418193655861545e-08,
+      "loss": 0.9286,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.675311150195928e-08,
+      "loss": 0.9322,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.609189272320237e-08,
+      "loss": 0.9665,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5434537576500856e-08,
+      "loss": 0.9196,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 0.8823,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.413141919940801e-08,
+      "loss": 0.9455,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.348565647533388e-08,
+      "loss": 0.9839,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.284375839594958e-08,
+      "loss": 0.9746,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.220572521066003e-08,
+      "loss": 0.989,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1571557167369165e-08,
+      "loss": 0.9476,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.094125451247656e-08,
+      "loss": 0.8578,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.031481749088296e-08,
+      "loss": 0.924,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9692246345985905e-08,
+      "loss": 0.8928,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9073541319680785e-08,
+      "loss": 1.0654,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8458702652359744e-08,
+      "loss": 0.9136,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.784773058291503e-08,
+      "loss": 0.9396,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7240625348735636e-08,
+      "loss": 0.9616,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.663738718570621e-08,
+      "loss": 0.9959,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.603801632821148e-08,
+      "loss": 0.8794,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.544251300913405e-08,
+      "loss": 0.9315,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.485087745984884e-08,
+      "loss": 0.9145,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4263109910234194e-08,
+      "loss": 0.9688,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.367921058866187e-08,
+      "loss": 0.9123,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3099179722002614e-08,
+      "loss": 0.9129,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.252301753562171e-08,
+      "loss": 0.8881,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.195072425338342e-08,
+      "loss": 0.9508,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1382300097647655e-08,
+      "loss": 0.9457,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0817745289272184e-08,
+      "loss": 0.97,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.025706004760932e-08,
+      "loss": 0.914,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.970024459051147e-08,
+      "loss": 0.8818,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.914729913432336e-08,
+      "loss": 0.9615,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.859822389388757e-08,
+      "loss": 0.9502,
+      "step": 5065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.805301908254455e-08,
+      "loss": 0.8944,
+      "step": 5066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.751168491212931e-08,
+      "loss": 0.9308,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.697422159297248e-08,
+      "loss": 0.9479,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6440629333901466e-08,
+      "loss": 0.8926,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.591090834224153e-08,
+      "loss": 0.9251,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.538505882380916e-08,
+      "loss": 0.9619,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4863080982919794e-08,
+      "loss": 0.93,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4344975022385654e-08,
+      "loss": 0.9483,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.383074114351237e-08,
+      "loss": 0.8962,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3320379546101234e-08,
+      "loss": 0.9228,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.281389042844918e-08,
+      "loss": 0.9822,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.23112739873499e-08,
+      "loss": 0.8984,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.181253041809052e-08,
+      "loss": 0.9782,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.13176599144549e-08,
+      "loss": 0.9499,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.082666266872036e-08,
+      "loss": 0.9715,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.033953887166097e-08,
+      "loss": 0.9286,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9856288712544204e-08,
+      "loss": 0.9356,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9376912379134314e-08,
+      "loss": 0.9067,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8901410057688982e-08,
+      "loss": 0.9088,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.842978193296153e-08,
+      "loss": 0.9536,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.796202818819871e-08,
+      "loss": 0.9145,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7498149005144025e-08,
+      "loss": 0.8656,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7038144564033307e-08,
+      "loss": 0.87,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.658201504359803e-08,
+      "loss": 0.9259,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6129760621063095e-08,
+      "loss": 1.0011,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5681381472151268e-08,
+      "loss": 0.9202,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5236877771074308e-08,
+      "loss": 0.8905,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.479624969054295e-08,
+      "loss": 0.8683,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4359497401758026e-08,
+      "loss": 0.9969,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3926621074417123e-08,
+      "loss": 0.8578,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.349762087671126e-08,
+      "loss": 0.8656,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3072496975325986e-08,
+      "loss": 0.9498,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.265124953543918e-08,
+      "loss": 0.9047,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2233878720723245e-08,
+      "loss": 0.941,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.182038469334513e-08,
+      "loss": 0.914,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1410767613965212e-08,
+      "loss": 0.9494,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1005027641736176e-08,
+      "loss": 0.9084,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0603164934306364e-08,
+      "loss": 0.9942,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020517964781532e-08,
+      "loss": 0.9329,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9811071936897132e-08,
+      "loss": 0.9751,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9420841954681525e-08,
+      "loss": 0.9576,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.903448985278722e-08,
+      "loss": 0.9032,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.86520157813308e-08,
+      "loss": 0.9313,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8273419888916733e-08,
+      "loss": 0.8841,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7898702322648453e-08,
+      "loss": 0.9805,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7527863228118393e-08,
+      "loss": 0.8845,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.716090274941351e-08,
+      "loss": 0.9322,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6797821029113093e-08,
+      "loss": 0.9129,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6438618208290957e-08,
+      "loss": 0.9591,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.608329442651213e-08,
+      "loss": 0.905,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5731849821833955e-08,
+      "loss": 0.9901,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5384284530810533e-08,
+      "loss": 0.8952,
+      "step": 5117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5040598688482732e-08,
+      "loss": 0.9407,
+      "step": 5118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4700792428388177e-08,
+      "loss": 0.9491,
+      "step": 5119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.436486588255681e-08,
+      "loss": 0.9449,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4032819181509783e-08,
+      "loss": 0.9641,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.370465245426167e-08,
+      "loss": 0.8763,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3380365828318254e-08,
+      "loss": 0.8807,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3059959429679859e-08,
+      "loss": 0.9981,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.274343338283801e-08,
+      "loss": 0.9268,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2430787810776556e-08,
+      "loss": 0.9641,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2122022834970548e-08,
+      "loss": 0.9387,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1817138575389576e-08,
+      "loss": 0.9602,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1516135150493323e-08,
+      "loss": 0.9556,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1219012677234908e-08,
+      "loss": 0.9289,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0925771271058649e-08,
+      "loss": 0.944,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0636411045902296e-08,
+      "loss": 0.92,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.035093211419369e-08,
+      "loss": 0.9597,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0069334586854106e-08,
+      "loss": 0.9171,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.791618573294914e-09,
+      "loss": 0.9517,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.517784181422018e-09,
+      "loss": 0.969,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.247831517632533e-09,
+      "loss": 0.9325,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.981760686811448e-09,
+      "loss": 0.9487,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.719571792341618e-09,
+      "loss": 0.9133,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.461264936093783e-09,
+      "loss": 0.912,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.20684021843099e-09,
+      "loss": 0.9119,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.956297738207496e-09,
+      "loss": 0.9315,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.70963759277099e-09,
+      "loss": 0.8811,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.466859877958143e-09,
+      "loss": 0.9163,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.227964688099054e-09,
+      "loss": 0.9151,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.992952116013918e-09,
+      "loss": 0.9428,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.76182225301747e-09,
+      "loss": 1.01,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.534575188911208e-09,
+      "loss": 0.9364,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3112110119900596e-09,
+      "loss": 0.8865,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.091729809042379e-09,
+      "loss": 0.9364,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.876131665345508e-09,
+      "loss": 0.9487,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.664416664666883e-09,
+      "loss": 0.894,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4565848892695894e-09,
+      "loss": 0.9485,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.252636419902368e-09,
+      "loss": 0.9198,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.052571335808499e-09,
+      "loss": 0.8966,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.856389714723575e-09,
+      "loss": 0.9203,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6640916328710705e-09,
+      "loss": 0.954,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.475677164966774e-09,
+      "loss": 0.8817,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2911463842176815e-09,
+      "loss": 0.929,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.110499362323106e-09,
+      "loss": 0.913,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.933736169471347e-09,
+      "loss": 0.9408,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.760856874341912e-09,
+      "loss": 0.9443,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.591861544106623e-09,
+      "loss": 0.9516,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4267502444274013e-09,
+      "loss": 0.9313,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2655230394562644e-09,
+      "loss": 0.8709,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1081799918375454e-09,
+      "loss": 0.9824,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.954721162704566e-09,
+      "loss": 0.9624,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.805146611684073e-09,
+      "loss": 1.0016,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6594563968918018e-09,
+      "loss": 0.9396,
+      "step": 5169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5176505749346937e-09,
+      "loss": 0.883,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.379729200908676e-09,
+      "loss": 0.9219,
+      "step": 5171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.245692328404214e-09,
+      "loss": 0.854,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.11554000949965e-09,
+      "loss": 0.9415,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9892722947645328e-09,
+      "loss": 0.9092,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8668892332596165e-09,
+      "loss": 0.9434,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7483908725357546e-09,
+      "loss": 0.9165,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6337772586338952e-09,
+      "loss": 0.9488,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5230484360873043e-09,
+      "loss": 0.9282,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4162044479182347e-09,
+      "loss": 0.9221,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3132453356412556e-09,
+      "loss": 0.82,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.214171139258813e-09,
+      "loss": 0.9311,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1189818972656697e-09,
+      "loss": 0.9352,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0276776466477955e-09,
+      "loss": 0.8854,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.40258422880147e-10,
+      "loss": 0.9333,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.567242599299974e-10,
+      "loss": 0.8406,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.770751902513862e-10,
+      "loss": 0.9124,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.013112447940007e-10,
+      "loss": 0.9651,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.294324529942942e-10,
+      "loss": 0.9013,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.614388427799266e-10,
+      "loss": 0.9202,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.973304405697654e-10,
+      "loss": 0.9556,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3710727127277417e-10,
+      "loss": 0.9524,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.807693582869032e-10,
+      "loss": 0.9331,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.283167235035301e-10,
+      "loss": 0.917,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.797493873019086e-10,
+      "loss": 0.9383,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.350673685536098e-10,
+      "loss": 0.9061,
+      "step": 5195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9427068461808086e-10,
+      "loss": 0.9903,
+      "step": 5196
+    },
+    {
+      "epoch": 1.0,
+      "step": 5196,
+      "total_flos": 6439427958505472.0,
+      "train_loss": 0.9808664602830467,
+      "train_runtime": 26027.9372,
+      "train_samples_per_second": 25.552,
+      "train_steps_per_second": 0.2
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5196,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 6439427958505472.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}