diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24990 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 39.71360381861575,
+  "eval_steps": 500,
+  "global_step": 4160,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2e-07,
+      "loss": 11.0204,
+      "step": 1
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4e-07,
+      "loss": 11.3347,
+      "step": 2
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 11.3476,
+      "step": 3
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.28e-06,
+      "loss": 11.3391,
+      "step": 4
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 11.267,
+      "step": 5
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 10.9396,
+      "step": 6
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.24e-06,
+      "loss": 10.7839,
+      "step": 7
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.56e-06,
+      "loss": 10.9837,
+      "step": 8
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.88e-06,
+      "loss": 11.5088,
+      "step": 9
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 11.3332,
+      "step": 10
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.52e-06,
+      "loss": 11.306,
+      "step": 11
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 11.3648,
+      "step": 12
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.16e-06,
+      "loss": 11.2193,
+      "step": 13
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.48e-06,
+      "loss": 11.3217,
+      "step": 14
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 11.2179,
+      "step": 15
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.12e-06,
+      "loss": 11.1602,
+      "step": 16
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 10.954,
+      "step": 17
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.76e-06,
+      "loss": 10.9509,
+      "step": 18
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.08e-06,
+      "loss": 10.8129,
+      "step": 19
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 11.018,
+      "step": 20
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 11.202,
+      "step": 21
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.04e-06,
+      "loss": 11.042,
+      "step": 22
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 11.1543,
+      "step": 23
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 11.1336,
+      "step": 24
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 10.8363,
+      "step": 25
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.32e-06,
+      "loss": 10.9331,
+      "step": 26
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64e-06,
+      "loss": 10.9245,
+      "step": 27
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.96e-06,
+      "loss": 10.9523,
+      "step": 28
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 10.914,
+      "step": 29
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 11.2724,
+      "step": 30
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 10.5219,
+      "step": 31
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.024e-05,
+      "loss": 10.2882,
+      "step": 32
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.056e-05,
+      "loss": 10.713,
+      "step": 33
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 10.3878,
+      "step": 34
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 10.6873,
+      "step": 35
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.152e-05,
+      "loss": 11.094,
+      "step": 36
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.184e-05,
+      "loss": 10.5487,
+      "step": 37
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.216e-05,
+      "loss": 10.7119,
+      "step": 38
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.248e-05,
+      "loss": 10.9098,
+      "step": 39
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 10.1222,
+      "step": 40
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 10.4132,
+      "step": 41
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 10.2507,
+      "step": 42
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.376e-05,
+      "loss": 10.4431,
+      "step": 43
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.408e-05,
+      "loss": 10.2475,
+      "step": 44
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 10.3885,
+      "step": 45
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 10.5804,
+      "step": 46
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 9.9755,
+      "step": 47
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 10.5727,
+      "step": 48
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 10.9149,
+      "step": 49
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 10.2608,
+      "step": 50
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.632e-05,
+      "loss": 10.3188,
+      "step": 51
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.664e-05,
+      "loss": 10.1081,
+      "step": 52
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.696e-05,
+      "loss": 10.2614,
+      "step": 53
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.728e-05,
+      "loss": 10.0735,
+      "step": 54
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.76e-05,
+      "loss": 9.9587,
+      "step": 55
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.792e-05,
+      "loss": 9.7899,
+      "step": 56
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.824e-05,
+      "loss": 9.7827,
+      "step": 57
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 9.437,
+      "step": 58
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 9.7448,
+      "step": 59
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 9.7051,
+      "step": 60
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 9.4802,
+      "step": 61
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 9.7981,
+      "step": 62
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.016e-05,
+      "loss": 9.1369,
+      "step": 63
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.048e-05,
+      "loss": 9.3745,
+      "step": 64
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.08e-05,
+      "loss": 9.4374,
+      "step": 65
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.112e-05,
+      "loss": 9.3749,
+      "step": 66
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.144e-05,
+      "loss": 8.9771,
+      "step": 67
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.1760000000000002e-05,
+      "loss": 9.182,
+      "step": 68
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.2080000000000002e-05,
+      "loss": 9.0217,
+      "step": 69
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 8.7273,
+      "step": 70
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.272e-05,
+      "loss": 9.3724,
+      "step": 71
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.304e-05,
+      "loss": 9.0563,
+      "step": 72
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.336e-05,
+      "loss": 8.7496,
+      "step": 73
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.368e-05,
+      "loss": 8.5054,
+      "step": 74
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4e-05,
+      "loss": 8.6348,
+      "step": 75
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.432e-05,
+      "loss": 8.722,
+      "step": 76
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.464e-05,
+      "loss": 8.4592,
+      "step": 77
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.496e-05,
+      "loss": 8.3022,
+      "step": 78
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5280000000000002e-05,
+      "loss": 8.1664,
+      "step": 79
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 8.065,
+      "step": 80
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5920000000000003e-05,
+      "loss": 8.2835,
+      "step": 81
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6240000000000003e-05,
+      "loss": 8.6761,
+      "step": 82
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6560000000000003e-05,
+      "loss": 8.1914,
+      "step": 83
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.6880000000000004e-05,
+      "loss": 7.5376,
+      "step": 84
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 8.0574,
+      "step": 85
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.752e-05,
+      "loss": 7.9332,
+      "step": 86
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.784e-05,
+      "loss": 7.9181,
+      "step": 87
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.816e-05,
+      "loss": 7.2633,
+      "step": 88
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8480000000000002e-05,
+      "loss": 8.0332,
+      "step": 89
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8800000000000002e-05,
+      "loss": 7.1838,
+      "step": 90
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9120000000000002e-05,
+      "loss": 7.491,
+      "step": 91
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9440000000000003e-05,
+      "loss": 7.3537,
+      "step": 92
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9760000000000003e-05,
+      "loss": 7.3017,
+      "step": 93
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0080000000000003e-05,
+      "loss": 7.2696,
+      "step": 94
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0400000000000004e-05,
+      "loss": 7.2796,
+      "step": 95
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0720000000000004e-05,
+      "loss": 7.1639,
+      "step": 96
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.104e-05,
+      "loss": 6.6451,
+      "step": 97
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1360000000000005e-05,
+      "loss": 6.8562,
+      "step": 98
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.168e-05,
+      "loss": 6.6974,
+      "step": 99
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 6.8668,
+      "step": 100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.232e-05,
+      "loss": 6.2042,
+      "step": 101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.264e-05,
+      "loss": 6.8482,
+      "step": 102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.296e-05,
+      "loss": 6.0705,
+      "step": 103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.328e-05,
+      "loss": 6.5763,
+      "step": 104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 6.0628,
+      "step": 105
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.392e-05,
+      "loss": 5.8228,
+      "step": 106
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4240000000000004e-05,
+      "loss": 5.6121,
+      "step": 107
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.456e-05,
+      "loss": 5.6801,
+      "step": 108
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.4880000000000005e-05,
+      "loss": 5.4932,
+      "step": 109
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.52e-05,
+      "loss": 5.6401,
+      "step": 110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.5520000000000006e-05,
+      "loss": 6.0535,
+      "step": 111
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.584e-05,
+      "loss": 5.3048,
+      "step": 112
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.6160000000000006e-05,
+      "loss": 5.0982,
+      "step": 113
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.648e-05,
+      "loss": 5.5283,
+      "step": 114
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.680000000000001e-05,
+      "loss": 5.1939,
+      "step": 115
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.7120000000000004e-05,
+      "loss": 4.8522,
+      "step": 116
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.744000000000001e-05,
+      "loss": 5.0927,
+      "step": 117
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7760000000000004e-05,
+      "loss": 5.0574,
+      "step": 118
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.808e-05,
+      "loss": 4.7663,
+      "step": 119
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 5.0567,
+      "step": 120
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.872e-05,
+      "loss": 4.4498,
+      "step": 121
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9040000000000006e-05,
+      "loss": 3.845,
+      "step": 122
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.936e-05,
+      "loss": 4.1361,
+      "step": 123
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9680000000000006e-05,
+      "loss": 3.9414,
+      "step": 124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4e-05,
+      "loss": 3.5702,
+      "step": 125
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.999999393804588e-05,
+      "loss": 4.1496,
+      "step": 126
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.999997575218719e-05,
+      "loss": 4.0599,
+      "step": 127
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.999994544243494e-05,
+      "loss": 3.1602,
+      "step": 128
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.999990300880752e-05,
+      "loss": 3.4558,
+      "step": 129
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.999984845133065e-05,
+      "loss": 3.9421,
+      "step": 130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9999781770037406e-05,
+      "loss": 3.0506,
+      "step": 131
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.99997029649682e-05,
+      "loss": 3.4821,
+      "step": 132
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.99996120361708e-05,
+      "loss": 3.7568,
+      "step": 133
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.999950898370035e-05,
+      "loss": 3.3811,
+      "step": 134
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9999393807619294e-05,
+      "loss": 3.3683,
+      "step": 135
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.999926650799746e-05,
+      "loss": 3.4777,
+      "step": 136
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.999912708491203e-05,
+      "loss": 3.4818,
+      "step": 137
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.999897553844751e-05,
+      "loss": 2.9197,
+      "step": 138
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.999881186869576e-05,
+      "loss": 2.775,
+      "step": 139
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.9998636075755996e-05,
+      "loss": 3.7353,
+      "step": 140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.99984481597348e-05,
+      "loss": 2.716,
+      "step": 141
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.9998248120746075e-05,
+      "loss": 2.747,
+      "step": 142
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.999803595891107e-05,
+      "loss": 2.726,
+      "step": 143
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.999781167435843e-05,
+      "loss": 2.6198,
+      "step": 144
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.999757526722409e-05,
+      "loss": 3.0415,
+      "step": 145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.9997326737651356e-05,
+      "loss": 2.8551,
+      "step": 146
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.99970660857909e-05,
+      "loss": 2.991,
+      "step": 147
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.9996793311800726e-05,
+      "loss": 3.0855,
+      "step": 148
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.999650841584618e-05,
+      "loss": 2.4385,
+      "step": 149
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.999621139809997e-05,
+      "loss": 2.8965,
+      "step": 150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.999590225874215e-05,
+      "loss": 2.675,
+      "step": 151
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.9995580997960116e-05,
+      "loss": 2.8499,
+      "step": 152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.9995247615948605e-05,
+      "loss": 2.3093,
+      "step": 153
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.999490211290973e-05,
+      "loss": 2.6816,
+      "step": 154
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.999454448905291e-05,
+      "loss": 2.7959,
+      "step": 155
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.9994174744594964e-05,
+      "loss": 2.4642,
+      "step": 156
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.999379287976e-05,
+      "loss": 2.9953,
+      "step": 157
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.999339889477953e-05,
+      "loss": 2.4893,
+      "step": 158
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.999299278989236e-05,
+      "loss": 2.6044,
+      "step": 159
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.999257456534469e-05,
+      "loss": 2.4149,
+      "step": 160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.999214422139003e-05,
+      "loss": 2.5528,
+      "step": 161
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.999170175828927e-05,
+      "loss": 2.456,
+      "step": 162
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.999124717631062e-05,
+      "loss": 2.4515,
+      "step": 163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.999078047572963e-05,
+      "loss": 3.3029,
+      "step": 164
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.999030165682924e-05,
+      "loss": 2.1499,
+      "step": 165
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.998981071989969e-05,
+      "loss": 2.6567,
+      "step": 166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.998930766523859e-05,
+      "loss": 2.2189,
+      "step": 167
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.998879249315087e-05,
+      "loss": 2.4112,
+      "step": 168
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9988265203948866e-05,
+      "loss": 2.15,
+      "step": 169
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9987725797952184e-05,
+      "loss": 2.4861,
+      "step": 170
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9987174275487814e-05,
+      "loss": 2.2275,
+      "step": 171
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.99866106368901e-05,
+      "loss": 2.3135,
+      "step": 172
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.998603488250071e-05,
+      "loss": 2.119,
+      "step": 173
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.998544701266866e-05,
+      "loss": 2.4417,
+      "step": 174
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.9984847027750316e-05,
+      "loss": 2.1587,
+      "step": 175
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.998423492810939e-05,
+      "loss": 2.1858,
+      "step": 176
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.998361071411693e-05,
+      "loss": 2.7351,
+      "step": 177
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.998297438615134e-05,
+      "loss": 2.2138,
+      "step": 178
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.998232594459834e-05,
+      "loss": 2.359,
+      "step": 179
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.998166538985103e-05,
+      "loss": 2.2938,
+      "step": 180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9980992722309826e-05,
+      "loss": 1.8524,
+      "step": 181
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9980307942382504e-05,
+      "loss": 2.0487,
+      "step": 182
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.997961105048417e-05,
+      "loss": 2.5457,
+      "step": 183
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9978902047037274e-05,
+      "loss": 2.3876,
+      "step": 184
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.997818093247162e-05,
+      "loss": 2.1367,
+      "step": 185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.997744770722433e-05,
+      "loss": 2.0066,
+      "step": 186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.997670237173989e-05,
+      "loss": 2.0063,
+      "step": 187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.997594492647012e-05,
+      "loss": 2.1464,
+      "step": 188
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.997517537187417e-05,
+      "loss": 2.3061,
+      "step": 189
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.9974393708418556e-05,
+      "loss": 1.9546,
+      "step": 190
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9973599936577113e-05,
+      "loss": 1.9901,
+      "step": 191
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.997279405683102e-05,
+      "loss": 2.1708,
+      "step": 192
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.9971976069668796e-05,
+      "loss": 2.1633,
+      "step": 193
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.99711459755863e-05,
+      "loss": 2.182,
+      "step": 194
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.997030377508674e-05,
+      "loss": 2.0662,
+      "step": 195
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.996944946868064e-05,
+      "loss": 1.8501,
+      "step": 196
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.9968583056885895e-05,
+      "loss": 2.2849,
+      "step": 197
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.9967704540227705e-05,
+      "loss": 2.0672,
+      "step": 198
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.996681391923863e-05,
+      "loss": 2.0908,
+      "step": 199
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.996591119445855e-05,
+      "loss": 2.0238,
+      "step": 200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.996499636643471e-05,
+      "loss": 2.2432,
+      "step": 201
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.996406943572166e-05,
+      "loss": 2.2882,
+      "step": 202
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.99631304028813e-05,
+      "loss": 2.0137,
+      "step": 203
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9962179268482886e-05,
+      "loss": 2.3796,
+      "step": 204
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.996121603310298e-05,
+      "loss": 2.0705,
+      "step": 205
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.996024069732547e-05,
+      "loss": 2.4449,
+      "step": 206
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.9959253261741644e-05,
+      "loss": 2.4706,
+      "step": 207
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.995825372695004e-05,
+      "loss": 1.9789,
+      "step": 208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.99572420935566e-05,
+      "loss": 2.0935,
+      "step": 209
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.995621836217455e-05,
+      "loss": 1.7726,
+      "step": 210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9955182533424485e-05,
+      "loss": 2.1901,
+      "step": 211
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995413460793431e-05,
+      "loss": 1.9201,
+      "step": 212
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9953074586339284e-05,
+      "loss": 2.2675,
+      "step": 213
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.995200246928198e-05,
+      "loss": 2.0339,
+      "step": 214
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.995091825741231e-05,
+      "loss": 2.1921,
+      "step": 215
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.994982195138753e-05,
+      "loss": 2.1918,
+      "step": 216
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9948713551872196e-05,
+      "loss": 2.136,
+      "step": 217
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.994759305953822e-05,
+      "loss": 1.9141,
+      "step": 218
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.994646047506485e-05,
+      "loss": 2.1534,
+      "step": 219
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.994531579913865e-05,
+      "loss": 2.4727,
+      "step": 220
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.994415903245351e-05,
+      "loss": 2.1083,
+      "step": 221
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.994299017571066e-05,
+      "loss": 1.6178,
+      "step": 222
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.994180922961867e-05,
+      "loss": 2.0044,
+      "step": 223
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9940616194893396e-05,
+      "loss": 1.9348,
+      "step": 224
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9939411072258066e-05,
+      "loss": 2.1713,
+      "step": 225
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9938193862443225e-05,
+      "loss": 1.9399,
+      "step": 226
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9936964566186735e-05,
+      "loss": 2.2397,
+      "step": 227
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.993572318423379e-05,
+      "loss": 1.853,
+      "step": 228
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.99344697173369e-05,
+      "loss": 2.1389,
+      "step": 229
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.993320416625592e-05,
+      "loss": 1.7419,
+      "step": 230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.993192653175803e-05,
+      "loss": 2.2572,
+      "step": 231
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.993063681461771e-05,
+      "loss": 1.7286,
+      "step": 232
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.992933501561679e-05,
+      "loss": 1.4725,
+      "step": 233
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.992802113554441e-05,
+      "loss": 1.9852,
+      "step": 234
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.992669517519705e-05,
+      "loss": 1.5181,
+      "step": 235
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.992535713537848e-05,
+      "loss": 2.2897,
+      "step": 236
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.992400701689983e-05,
+      "loss": 1.8499,
+      "step": 237
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.992264482057953e-05,
+      "loss": 1.6927,
+      "step": 238
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.992127054724333e-05,
+      "loss": 2.336,
+      "step": 239
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.991988419772433e-05,
+      "loss": 1.7773,
+      "step": 240
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.99184857728629e-05,
+      "loss": 1.9823,
+      "step": 241
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.991707527350679e-05,
+      "loss": 1.9172,
+      "step": 242
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.991565270051101e-05,
+      "loss": 2.4358,
+      "step": 243
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.991421805473793e-05,
+      "loss": 2.5147,
+      "step": 244
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.991277133705723e-05,
+      "loss": 1.5019,
+      "step": 245
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.991131254834589e-05,
+      "loss": 2.3903,
+      "step": 246
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9909841689488235e-05,
+      "loss": 2.6465,
+      "step": 247
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.990835876137589e-05,
+      "loss": 1.8536,
+      "step": 248
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.990686376490779e-05,
+      "loss": 2.3167,
+      "step": 249
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.990535670099021e-05,
+      "loss": 1.953,
+      "step": 250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9903837570536704e-05,
+      "loss": 1.8831,
+      "step": 251
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9902306374468186e-05,
+      "loss": 1.6854,
+      "step": 252
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.990076311371284e-05,
+      "loss": 2.2017,
+      "step": 253
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9899207789206196e-05,
+      "loss": 2.0832,
+      "step": 254
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.989764040189108e-05,
+      "loss": 1.8489,
+      "step": 255
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.989606095271764e-05,
+      "loss": 1.8991,
+      "step": 256
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.989446944264332e-05,
+      "loss": 2.1151,
+      "step": 257
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.989286587263289e-05,
+      "loss": 1.7424,
+      "step": 258
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.989125024365844e-05,
+      "loss": 1.5999,
+      "step": 259
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.988962255669933e-05,
+      "loss": 1.8225,
+      "step": 260
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.9887982812742295e-05,
+      "loss": 1.8642,
+      "step": 261
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.98863310127813e-05,
+      "loss": 2.095,
+      "step": 262
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.9884667157817685e-05,
+      "loss": 1.9813,
+      "step": 263
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.988299124886006e-05,
+      "loss": 2.0172,
+      "step": 264
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.9881303286924356e-05,
+      "loss": 2.1506,
+      "step": 265
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.987960327303381e-05,
+      "loss": 1.4285,
+      "step": 266
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.9877891208218956e-05,
+      "loss": 1.7106,
+      "step": 267
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.9876167093517654e-05,
+      "loss": 1.834,
+      "step": 268
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.987443092997503e-05,
+      "loss": 2.261,
+      "step": 269
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.987268271864356e-05,
+      "loss": 2.3845,
+      "step": 270
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.9870922460583e-05,
+      "loss": 1.8182,
+      "step": 271
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.98691501568604e-05,
+      "loss": 1.3387,
+      "step": 272
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.986736580855013e-05,
+      "loss": 2.1228,
+      "step": 273
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.986556941673385e-05,
+      "loss": 2.0708,
+      "step": 274
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.9863760982500526e-05,
+      "loss": 1.5791,
+      "step": 275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.986194050694642e-05,
+      "loss": 2.2648,
+      "step": 276
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.98601079911751e-05,
+      "loss": 1.2487,
+      "step": 277
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.985826343629743e-05,
+      "loss": 1.9657,
+      "step": 278
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.9856406843431556e-05,
+      "loss": 1.881,
+      "step": 279
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.9854538213702963e-05,
+      "loss": 2.0767,
+      "step": 280
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.985265754824438e-05,
+      "loss": 1.7936,
+      "step": 281
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.985076484819588e-05,
+      "loss": 1.9673,
+      "step": 282
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.984886011470478e-05,
+      "loss": 1.6774,
+      "step": 283
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.984694334892575e-05,
+      "loss": 2.1916,
+      "step": 284
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.984501455202071e-05,
+      "loss": 1.5492,
+      "step": 285
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.984307372515889e-05,
+      "loss": 1.7264,
+      "step": 286
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9841120869516815e-05,
+      "loss": 2.0507,
+      "step": 287
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.9839155986278286e-05,
+      "loss": 1.9751,
+      "step": 288
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.983717907663441e-05,
+      "loss": 2.1079,
+      "step": 289
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.983519014178359e-05,
+      "loss": 1.7926,
+      "step": 290
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.983318918293151e-05,
+      "loss": 2.0623,
+      "step": 291
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9831176201291126e-05,
+      "loss": 2.0077,
+      "step": 292
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.982915119808271e-05,
+      "loss": 1.8915,
+      "step": 293
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9827114174533804e-05,
+      "loss": 1.7056,
+      "step": 294
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.9825065131879246e-05,
+      "loss": 2.0389,
+      "step": 295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.9823004071361156e-05,
+      "loss": 1.8437,
+      "step": 296
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.982093099422894e-05,
+      "loss": 1.7038,
+      "step": 297
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.9818845901739284e-05,
+      "loss": 1.85,
+      "step": 298
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.981674879515616e-05,
+      "loss": 1.8053,
+      "step": 299
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.981463967575084e-05,
+      "loss": 1.5789,
+      "step": 300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.981251854480185e-05,
+      "loss": 1.5857,
+      "step": 301
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.9810385403595004e-05,
+      "loss": 1.9362,
+      "step": 302
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.9808240253423414e-05,
+      "loss": 1.8946,
+      "step": 303
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9806083095587456e-05,
+      "loss": 1.4918,
+      "step": 304
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.9803913931394775e-05,
+      "loss": 2.2044,
+      "step": 305
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9801732762160344e-05,
+      "loss": 1.8376,
+      "step": 306
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.979953958920634e-05,
+      "loss": 1.5853,
+      "step": 307
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9797334413862276e-05,
+      "loss": 2.1572,
+      "step": 308
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.979511723746491e-05,
+      "loss": 1.8776,
+      "step": 309
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.979288806135829e-05,
+      "loss": 2.0389,
+      "step": 310
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.979064688689374e-05,
+      "loss": 1.7352,
+      "step": 311
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.9788393715429825e-05,
+      "loss": 1.7567,
+      "step": 312
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.978612854833243e-05,
+      "loss": 2.0867,
+      "step": 313
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9783851386974676e-05,
+      "loss": 1.8773,
+      "step": 314
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.978156223273697e-05,
+      "loss": 1.5591,
+      "step": 315
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9779261087006994e-05,
+      "loss": 1.6834,
+      "step": 316
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.977694795117969e-05,
+      "loss": 1.7154,
+      "step": 317
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.977462282665726e-05,
+      "loss": 1.8139,
+      "step": 318
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.97722857148492e-05,
+      "loss": 1.6217,
+      "step": 319
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.976993661717224e-05,
+      "loss": 2.0896,
+      "step": 320
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.9767575535050404e-05,
+      "loss": 1.4903,
+      "step": 321
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.976520246991497e-05,
+      "loss": 1.7126,
+      "step": 322
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9762817423204465e-05,
+      "loss": 1.7513,
+      "step": 323
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.976042039636471e-05,
+      "loss": 2.0055,
+      "step": 324
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.975801139084876e-05,
+      "loss": 1.7137,
+      "step": 325
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.9755590408116956e-05,
+      "loss": 1.8467,
+      "step": 326
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.975315744963687e-05,
+      "loss": 2.3026,
+      "step": 327
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.975071251688336e-05,
+      "loss": 1.5341,
+      "step": 328
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.974825561133853e-05,
+      "loss": 1.7031,
+      "step": 329
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.974578673449175e-05,
+      "loss": 1.4174,
+      "step": 330
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.974330588783964e-05,
+      "loss": 1.7804,
+      "step": 331
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.974081307288607e-05,
+      "loss": 1.9574,
+      "step": 332
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.973830829114218e-05,
+      "loss": 1.8439,
+      "step": 333
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.973579154412636e-05,
+      "loss": 2.0215,
+      "step": 334
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.973326283336424e-05,
+      "loss": 1.7691,
+      "step": 335
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.973072216038873e-05,
+      "loss": 1.7592,
+      "step": 336
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.972816952673995e-05,
+      "loss": 1.8082,
+      "step": 337
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.972560493396532e-05,
+      "loss": 1.5441,
+      "step": 338
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9723028383619476e-05,
+      "loss": 1.8546,
+      "step": 339
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.97204398772643e-05,
+      "loss": 1.8961,
+      "step": 340
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.9717839416468944e-05,
+      "loss": 1.7397,
+      "step": 341
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.971522700280979e-05,
+      "loss": 1.5469,
+      "step": 342
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9712602637870475e-05,
+      "loss": 1.6812,
+      "step": 343
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.970996632324187e-05,
+      "loss": 1.7073,
+      "step": 344
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.970731806052211e-05,
+      "loss": 1.5709,
+      "step": 345
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.970465785131655e-05,
+      "loss": 2.1812,
+      "step": 346
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.970198569723779e-05,
+      "loss": 2.0034,
+      "step": 347
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.96993015999057e-05,
+      "loss": 1.6972,
+      "step": 348
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.969660556094735e-05,
+      "loss": 1.5039,
+      "step": 349
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.9693897581997064e-05,
+      "loss": 1.6024,
+      "step": 350
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.9691177664696414e-05,
+      "loss": 1.8556,
+      "step": 351
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.9688445810694205e-05,
+      "loss": 1.395,
+      "step": 352
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.9685702021646466e-05,
+      "loss": 1.7734,
+      "step": 353
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.968294629921647e-05,
+      "loss": 1.79,
+      "step": 354
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.9680178645074726e-05,
+      "loss": 1.9131,
+      "step": 355
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.9677399060898977e-05,
+      "loss": 1.69,
+      "step": 356
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.9674607548374186e-05,
+      "loss": 2.0688,
+      "step": 357
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.967180410919256e-05,
+      "loss": 2.4486,
+      "step": 358
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.966898874505353e-05,
+      "loss": 1.9025,
+      "step": 359
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.966616145766376e-05,
+      "loss": 1.6186,
+      "step": 360
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.9663322248737144e-05,
+      "loss": 1.7347,
+      "step": 361
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.966047111999477e-05,
+      "loss": 1.9125,
+      "step": 362
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.965760807316501e-05,
+      "loss": 2.104,
+      "step": 363
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.965473310998342e-05,
+      "loss": 2.0794,
+      "step": 364
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.965184623219278e-05,
+      "loss": 1.8146,
+      "step": 365
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.964894744154312e-05,
+      "loss": 1.9238,
+      "step": 366
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.9646036739791656e-05,
+      "loss": 1.4239,
+      "step": 367
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.9643114128702846e-05,
+      "loss": 1.6719,
+      "step": 368
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.964017961004837e-05,
+      "loss": 1.7618,
+      "step": 369
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.963723318560712e-05,
+      "loss": 1.6016,
+      "step": 370
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.96342748571652e-05,
+      "loss": 1.9785,
+      "step": 371
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.9631304626515935e-05,
+      "loss": 1.6969,
+      "step": 372
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.962832249545986e-05,
+      "loss": 2.2888,
+      "step": 373
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.962532846580475e-05,
+      "loss": 1.8465,
+      "step": 374
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.962232253936555e-05,
+      "loss": 1.9871,
+      "step": 375
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.961930471796446e-05,
+      "loss": 1.7766,
+      "step": 376
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.9616275003430836e-05,
+      "loss": 1.6498,
+      "step": 377
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.961323339760131e-05,
+      "loss": 2.0121,
+      "step": 378
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.961017990231967e-05,
+      "loss": 1.463,
+      "step": 379
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.9607114519436945e-05,
+      "loss": 1.5161,
+      "step": 380
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.9604037250811345e-05,
+      "loss": 1.7031,
+      "step": 381
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.96009480983083e-05,
+      "loss": 2.0301,
+      "step": 382
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.959784706380044e-05,
+      "loss": 1.6469,
+      "step": 383
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.95947341491676e-05,
+      "loss": 1.8564,
+      "step": 384
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.9591609356296806e-05,
+      "loss": 1.9718,
+      "step": 385
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.958847268708231e-05,
+      "loss": 1.927,
+      "step": 386
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.958532414342553e-05,
+      "loss": 1.2901,
+      "step": 387
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.95821637272351e-05,
+      "loss": 1.4833,
+      "step": 388
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.957899144042686e-05,
+      "loss": 1.4964,
+      "step": 389
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.9575807284923825e-05,
+      "loss": 1.666,
+      "step": 390
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.957261126265623e-05,
+      "loss": 1.715,
+      "step": 391
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.9569403375561475e-05,
+      "loss": 1.8609,
+      "step": 392
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.956618362558417e-05,
+      "loss": 1.755,
+      "step": 393
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.9562952014676116e-05,
+      "loss": 1.8285,
+      "step": 394
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.95597085447963e-05,
+      "loss": 1.7679,
+      "step": 395
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.955645321791089e-05,
+      "loss": 1.8782,
+      "step": 396
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.9553186035993263e-05,
+      "loss": 1.8486,
+      "step": 397
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.9549907001023966e-05,
+      "loss": 1.7391,
+      "step": 398
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.954661611499073e-05,
+      "loss": 2.0823,
+      "step": 399
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9543313379888477e-05,
+      "loss": 1.5872,
+      "step": 400
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.953999879771932e-05,
+      "loss": 1.5674,
+      "step": 401
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.9536672370492517e-05,
+      "loss": 2.035,
+      "step": 402
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.9533334100224566e-05,
+      "loss": 1.5493,
+      "step": 403
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.9529983988939083e-05,
+      "loss": 1.5652,
+      "step": 404
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.9526622038666914e-05,
+      "loss": 1.8049,
+      "step": 405
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.9523248251446046e-05,
+      "loss": 1.7329,
+      "step": 406
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.9519862629321645e-05,
+      "loss": 1.7267,
+      "step": 407
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.951646517434607e-05,
+      "loss": 1.5692,
+      "step": 408
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.9513055888578846e-05,
+      "loss": 2.2458,
+      "step": 409
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.950963477408666e-05,
+      "loss": 1.9181,
+      "step": 410
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.9506201832943375e-05,
+      "loss": 2.1104,
+      "step": 411
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.950275706723003e-05,
+      "loss": 1.8898,
+      "step": 412
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.949930047903482e-05,
+      "loss": 1.491,
+      "step": 413
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.9495832070453114e-05,
+      "loss": 1.6082,
+      "step": 414
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.949235184358744e-05,
+      "loss": 2.0418,
+      "step": 415
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.9488859800547523e-05,
+      "loss": 1.4257,
+      "step": 416
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.948535594345018e-05,
+      "loss": 1.6657,
+      "step": 417
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.948184027441947e-05,
+      "loss": 1.7836,
+      "step": 418
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.947831279558656e-05,
+      "loss": 2.0988,
+      "step": 419
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.947477350908979e-05,
+      "loss": 1.4794,
+      "step": 420
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.947122241707466e-05,
+      "loss": 1.4261,
+      "step": 421
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9467659521693836e-05,
+      "loss": 1.4691,
+      "step": 422
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.9464084825107117e-05,
+      "loss": 1.7381,
+      "step": 423
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.9460498329481465e-05,
+      "loss": 1.7072,
+      "step": 424
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.945690003699101e-05,
+      "loss": 1.6691,
+      "step": 425
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.9453289949817014e-05,
+      "loss": 1.9834,
+      "step": 426
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.9449668070147896e-05,
+      "loss": 1.6453,
+      "step": 427
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.944603440017922e-05,
+      "loss": 1.4294,
+      "step": 428
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.94423889421137e-05,
+      "loss": 1.6129,
+      "step": 429
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.94387316981612e-05,
+      "loss": 1.5778,
+      "step": 430
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.943506267053872e-05,
+      "loss": 1.9607,
+      "step": 431
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.943138186147041e-05,
+      "loss": 1.5905,
+      "step": 432
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.9427689273187565e-05,
+      "loss": 1.8445,
+      "step": 433
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.942398490792861e-05,
+      "loss": 1.4429,
+      "step": 434
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.942026876793911e-05,
+      "loss": 1.7187,
+      "step": 435
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.941654085547178e-05,
+      "loss": 1.5983,
+      "step": 436
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.9412801172786453e-05,
+      "loss": 1.4323,
+      "step": 437
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.9409049722150125e-05,
+      "loss": 1.8642,
+      "step": 438
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.940528650583689e-05,
+      "loss": 1.6148,
+      "step": 439
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.9401511526128e-05,
+      "loss": 1.6024,
+      "step": 440
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.939772478531183e-05,
+      "loss": 1.6544,
+      "step": 441
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.9393926285683884e-05,
+      "loss": 1.8514,
+      "step": 442
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.9390116029546804e-05,
+      "loss": 1.9234,
+      "step": 443
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.938629401921033e-05,
+      "loss": 1.7171,
+      "step": 444
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.938246025699137e-05,
+      "loss": 1.4425,
+      "step": 445
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.937861474521391e-05,
+      "loss": 2.0304,
+      "step": 446
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.9374757486209106e-05,
+      "loss": 2.0302,
+      "step": 447
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.937088848231519e-05,
+      "loss": 1.8013,
+      "step": 448
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.936700773587754e-05,
+      "loss": 1.8098,
+      "step": 449
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.936311524924866e-05,
+      "loss": 2.215,
+      "step": 450
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.935921102478814e-05,
+      "loss": 1.4868,
+      "step": 451
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.935529506486272e-05,
+      "loss": 1.5864,
+      "step": 452
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.935136737184621e-05,
+      "loss": 1.7109,
+      "step": 453
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.934742794811959e-05,
+      "loss": 1.7265,
+      "step": 454
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.93434767960709e-05,
+      "loss": 1.7087,
+      "step": 455
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.933951391809532e-05,
+      "loss": 1.9094,
+      "step": 456
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.933553931659512e-05,
+      "loss": 1.8207,
+      "step": 457
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.93315529939797e-05,
+      "loss": 1.8693,
+      "step": 458
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.932755495266554e-05,
+      "loss": 1.5265,
+      "step": 459
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.932354519507623e-05,
+      "loss": 1.6438,
+      "step": 460
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.931952372364248e-05,
+      "loss": 1.4404,
+      "step": 461
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.9315490540802076e-05,
+      "loss": 1.7392,
+      "step": 462
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.931144564899992e-05,
+      "loss": 1.7787,
+      "step": 463
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.9307389050688e-05,
+      "loss": 1.6277,
+      "step": 464
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.9303320748325415e-05,
+      "loss": 1.8087,
+      "step": 465
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.9299240744378355e-05,
+      "loss": 1.5143,
+      "step": 466
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.9295149041320087e-05,
+      "loss": 1.7809,
+      "step": 467
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.929104564163099e-05,
+      "loss": 1.4689,
+      "step": 468
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.928693054779853e-05,
+      "loss": 1.6673,
+      "step": 469
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.928280376231725e-05,
+      "loss": 1.7562,
+      "step": 470
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.9278665287688796e-05,
+      "loss": 1.5652,
+      "step": 471
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.927451512642188e-05,
+      "loss": 2.0781,
+      "step": 472
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.927035328103233e-05,
+      "loss": 1.3862,
+      "step": 473
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.926617975404302e-05,
+      "loss": 1.9441,
+      "step": 474
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.926199454798393e-05,
+      "loss": 2.1527,
+      "step": 475
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.925779766539211e-05,
+      "loss": 1.8326,
+      "step": 476
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.92535891088117e-05,
+      "loss": 1.8974,
+      "step": 477
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9249368880793895e-05,
+      "loss": 1.4333,
+      "step": 478
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9245136983896985e-05,
+      "loss": 1.5203,
+      "step": 479
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.9240893420686326e-05,
+      "loss": 1.5529,
+      "step": 480
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.9236638193734345e-05,
+      "loss": 1.3736,
+      "step": 481
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.923237130562054e-05,
+      "loss": 1.3565,
+      "step": 482
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.922809275893149e-05,
+      "loss": 2.0386,
+      "step": 483
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.922380255626081e-05,
+      "loss": 1.7426,
+      "step": 484
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.921950070020922e-05,
+      "loss": 1.85,
+      "step": 485
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.9215187193384474e-05,
+      "loss": 1.5996,
+      "step": 486
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.921086203840141e-05,
+      "loss": 2.0808,
+      "step": 487
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.92065252378819e-05,
+      "loss": 2.1025,
+      "step": 488
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.9202176794454905e-05,
+      "loss": 1.4663,
+      "step": 489
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.9197816710756435e-05,
+      "loss": 1.5204,
+      "step": 490
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.9193444989429543e-05,
+      "loss": 1.888,
+      "step": 491
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.918906163312436e-05,
+      "loss": 1.9221,
+      "step": 492
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.9184666644498043e-05,
+      "loss": 1.6384,
+      "step": 493
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.9180260026214815e-05,
+      "loss": 1.86,
+      "step": 494
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.917584178094596e-05,
+      "loss": 1.4377,
+      "step": 495
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.9171411911369774e-05,
+      "loss": 1.5318,
+      "step": 496
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.916697042017165e-05,
+      "loss": 1.8171,
+      "step": 497
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.916251731004399e-05,
+      "loss": 1.3452,
+      "step": 498
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.9158052583686246e-05,
+      "loss": 1.2326,
+      "step": 499
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.9153576243804914e-05,
+      "loss": 1.5986,
+      "step": 500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.914908829311353e-05,
+      "loss": 1.6274,
+      "step": 501
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.914458873433268e-05,
+      "loss": 1.7552,
+      "step": 502
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.914007757018996e-05,
+      "loss": 1.4381,
+      "step": 503
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.913555480342003e-05,
+      "loss": 2.1199,
+      "step": 504
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.913102043676457e-05,
+      "loss": 1.556,
+      "step": 505
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.9126474472972276e-05,
+      "loss": 2.0567,
+      "step": 506
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.91219169147989e-05,
+      "loss": 1.7837,
+      "step": 507
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.9117347765007226e-05,
+      "loss": 1.7057,
+      "step": 508
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.911276702636704e-05,
+      "loss": 1.71,
+      "step": 509
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.910817470165516e-05,
+      "loss": 1.7732,
+      "step": 510
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.910357079365544e-05,
+      "loss": 1.8421,
+      "step": 511
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.909895530515874e-05,
+      "loss": 1.6655,
+      "step": 512
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.909432823896296e-05,
+      "loss": 1.9355,
+      "step": 513
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.9089689597873e-05,
+      "loss": 1.5737,
+      "step": 514
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.908503938470078e-05,
+      "loss": 1.4371,
+      "step": 515
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.908037760226524e-05,
+      "loss": 1.9046,
+      "step": 516
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.9075704253392326e-05,
+      "loss": 1.7692,
+      "step": 517
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.907101934091501e-05,
+      "loss": 1.626,
+      "step": 518
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.9066322867673266e-05,
+      "loss": 1.8668,
+      "step": 519
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.906161483651406e-05,
+      "loss": 1.5294,
+      "step": 520
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.90568952502914e-05,
+      "loss": 1.5679,
+      "step": 521
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.905216411186626e-05,
+      "loss": 1.3218,
+      "step": 522
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.904742142410664e-05,
+      "loss": 1.8378,
+      "step": 523
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.904266718988753e-05,
+      "loss": 1.9363,
+      "step": 524
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.903790141209094e-05,
+      "loss": 1.6177,
+      "step": 525
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.9033124093605844e-05,
+      "loss": 1.2626,
+      "step": 526
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.902833523732824e-05,
+      "loss": 1.6034,
+      "step": 527
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.9023534846161116e-05,
+      "loss": 1.4697,
+      "step": 528
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.901872292301444e-05,
+      "loss": 1.4627,
+      "step": 529
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.901389947080518e-05,
+      "loss": 1.5734,
+      "step": 530
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.900906449245729e-05,
+      "loss": 1.61,
+      "step": 531
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.900421799090171e-05,
+      "loss": 1.5946,
+      "step": 532
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.899935996907637e-05,
+      "loss": 1.7766,
+      "step": 533
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.8994490429926175e-05,
+      "loss": 1.9666,
+      "step": 534
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.8989609376403025e-05,
+      "loss": 1.5169,
+      "step": 535
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.8984716811465787e-05,
+      "loss": 1.6742,
+      "step": 536
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.8979812738080314e-05,
+      "loss": 1.4184,
+      "step": 537
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.897489715921943e-05,
+      "loss": 1.805,
+      "step": 538
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.8969970077862934e-05,
+      "loss": 1.3572,
+      "step": 539
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.8965031496997614e-05,
+      "loss": 1.5531,
+      "step": 540
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.89600814196172e-05,
+      "loss": 1.4074,
+      "step": 541
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.895511984872241e-05,
+      "loss": 1.1467,
+      "step": 542
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.895014678732094e-05,
+      "loss": 1.7523,
+      "step": 543
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.8945162238427405e-05,
+      "loss": 1.9978,
+      "step": 544
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.894016620506345e-05,
+      "loss": 1.6688,
+      "step": 545
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.893515869025763e-05,
+      "loss": 1.4248,
+      "step": 546
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.893013969704547e-05,
+      "loss": 2.0494,
+      "step": 547
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.8925109228469474e-05,
+      "loss": 2.2122,
+      "step": 548
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.892006728757909e-05,
+      "loss": 1.7371,
+      "step": 549
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.8915013877430715e-05,
+      "loss": 1.9236,
+      "step": 550
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.89099490010877e-05,
+      "loss": 2.2029,
+      "step": 551
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.890487266162035e-05,
+      "loss": 1.6349,
+      "step": 552
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.8899784862105925e-05,
+      "loss": 1.6803,
+      "step": 553
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.889468560562862e-05,
+      "loss": 1.5485,
+      "step": 554
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.888957489527959e-05,
+      "loss": 1.565,
+      "step": 555
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.88844527341569e-05,
+      "loss": 1.6284,
+      "step": 556
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.887931912536561e-05,
+      "loss": 1.5534,
+      "step": 557
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.8874174072017686e-05,
+      "loss": 1.8338,
+      "step": 558
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.886901757723202e-05,
+      "loss": 1.3822,
+      "step": 559
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.886384964413446e-05,
+      "loss": 1.7229,
+      "step": 560
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.8858670275857786e-05,
+      "loss": 1.5323,
+      "step": 561
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.885347947554171e-05,
+      "loss": 1.8106,
+      "step": 562
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.8848277246332864e-05,
+      "loss": 2.1027,
+      "step": 563
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.884306359138483e-05,
+      "loss": 1.8797,
+      "step": 564
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.883783851385809e-05,
+      "loss": 1.5573,
+      "step": 565
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.8832602016920056e-05,
+      "loss": 1.823,
+      "step": 566
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.8827354103745084e-05,
+      "loss": 1.8139,
+      "step": 567
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.8822094777514425e-05,
+      "loss": 1.6306,
+      "step": 568
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.881682404141626e-05,
+      "loss": 1.7705,
+      "step": 569
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.881154189864569e-05,
+      "loss": 1.6586,
+      "step": 570
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.8806248352404716e-05,
+      "loss": 1.8307,
+      "step": 571
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.8800943405902275e-05,
+      "loss": 1.9422,
+      "step": 572
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.879562706235419e-05,
+      "loss": 1.7019,
+      "step": 573
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.879029932498321e-05,
+      "loss": 1.6966,
+      "step": 574
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.8784960197018975e-05,
+      "loss": 1.6619,
+      "step": 575
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.877960968169805e-05,
+      "loss": 1.6068,
+      "step": 576
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.8774247782263896e-05,
+      "loss": 1.5676,
+      "step": 577
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.876887450196686e-05,
+      "loss": 1.5407,
+      "step": 578
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.876348984406421e-05,
+      "loss": 1.5548,
+      "step": 579
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.875809381182009e-05,
+      "loss": 1.5103,
+      "step": 580
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.875268640850556e-05,
+      "loss": 1.4305,
+      "step": 581
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.8747267637398565e-05,
+      "loss": 1.6372,
+      "step": 582
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.874183750178393e-05,
+      "loss": 1.7128,
+      "step": 583
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.873639600495338e-05,
+      "loss": 1.7868,
+      "step": 584
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.873094315020553e-05,
+      "loss": 2.2861,
+      "step": 585
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.872547894084587e-05,
+      "loss": 1.6198,
+      "step": 586
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.8720003380186785e-05,
+      "loss": 1.6445,
+      "step": 587
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.871451647154754e-05,
+      "loss": 1.7591,
+      "step": 588
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.870901821825425e-05,
+      "loss": 1.939,
+      "step": 589
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.8703508623639956e-05,
+      "loss": 1.4723,
+      "step": 590
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.869798769104454e-05,
+      "loss": 1.5069,
+      "step": 591
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.869245542381476e-05,
+      "loss": 1.6768,
+      "step": 592
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.8686911825304256e-05,
+      "loss": 1.3958,
+      "step": 593
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.868135689887354e-05,
+      "loss": 1.6126,
+      "step": 594
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.867579064788997e-05,
+      "loss": 1.3658,
+      "step": 595
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.8670213075727785e-05,
+      "loss": 1.7524,
+      "step": 596
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.866462418576808e-05,
+      "loss": 1.4704,
+      "step": 597
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.865902398139884e-05,
+      "loss": 1.7655,
+      "step": 598
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.865341246601485e-05,
+      "loss": 1.8814,
+      "step": 599
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.86477896430178e-05,
+      "loss": 1.9063,
+      "step": 600
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.8642155515816215e-05,
+      "loss": 1.3224,
+      "step": 601
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.863651008782549e-05,
+      "loss": 1.4207,
+      "step": 602
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.8630853362467836e-05,
+      "loss": 1.9325,
+      "step": 603
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.8625185343172346e-05,
+      "loss": 1.5466,
+      "step": 604
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.8619506033374956e-05,
+      "loss": 1.8435,
+      "step": 605
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.861381543651843e-05,
+      "loss": 1.722,
+      "step": 606
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.8608113556052365e-05,
+      "loss": 1.859,
+      "step": 607
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.860240039543324e-05,
+      "loss": 1.6801,
+      "step": 608
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.8596675958124335e-05,
+      "loss": 2.0657,
+      "step": 609
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.8590940247595773e-05,
+      "loss": 1.2541,
+      "step": 610
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.858519326732453e-05,
+      "loss": 1.8487,
+      "step": 611
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.8579435020794396e-05,
+      "loss": 1.395,
+      "step": 612
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.8573665511495974e-05,
+      "loss": 2.0131,
+      "step": 613
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.856788474292674e-05,
+      "loss": 1.7632,
+      "step": 614
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.8562092718590944e-05,
+      "loss": 1.7766,
+      "step": 615
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.8556289441999714e-05,
+      "loss": 1.3067,
+      "step": 616
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.855047491667094e-05,
+      "loss": 1.6348,
+      "step": 617
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.854464914612938e-05,
+      "loss": 1.5188,
+      "step": 618
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.853881213390659e-05,
+      "loss": 1.8637,
+      "step": 619
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.8532963883540925e-05,
+      "loss": 1.5087,
+      "step": 620
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.8527104398577584e-05,
+      "loss": 1.9367,
+      "step": 621
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.852123368256854e-05,
+      "loss": 1.7766,
+      "step": 622
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.851535173907262e-05,
+      "loss": 1.458,
+      "step": 623
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.8509458571655414e-05,
+      "loss": 1.7696,
+      "step": 624
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.850355418388933e-05,
+      "loss": 1.7072,
+      "step": 625
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.8497638579353595e-05,
+      "loss": 1.2364,
+      "step": 626
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.849171176163421e-05,
+      "loss": 1.888,
+      "step": 627
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.8485773734323995e-05,
+      "loss": 1.5696,
+      "step": 628
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.8479824501022546e-05,
+      "loss": 1.6914,
+      "step": 629
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.847386406533626e-05,
+      "loss": 1.8924,
+      "step": 630
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.8467892430878325e-05,
+      "loss": 1.653,
+      "step": 631
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.846190960126873e-05,
+      "loss": 1.4109,
+      "step": 632
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.8455915580134234e-05,
+      "loss": 1.5014,
+      "step": 633
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.844991037110837e-05,
+      "loss": 1.4131,
+      "step": 634
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.844389397783149e-05,
+      "loss": 1.9975,
+      "step": 635
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.84378664039507e-05,
+      "loss": 1.9176,
+      "step": 636
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.843182765311987e-05,
+      "loss": 1.5607,
+      "step": 637
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.8425777728999676e-05,
+      "loss": 1.6071,
+      "step": 638
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.841971663525756e-05,
+      "loss": 1.8193,
+      "step": 639
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.841364437556772e-05,
+      "loss": 1.4338,
+      "step": 640
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.840756095361114e-05,
+      "loss": 2.1033,
+      "step": 641
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.8401466373075555e-05,
+      "loss": 1.6929,
+      "step": 642
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.839536063765547e-05,
+      "loss": 1.8756,
+      "step": 643
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.8389243751052153e-05,
+      "loss": 1.8183,
+      "step": 644
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.838311571697365e-05,
+      "loss": 1.8925,
+      "step": 645
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.8376976539134726e-05,
+      "loss": 1.5247,
+      "step": 646
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.837082622125694e-05,
+      "loss": 1.8047,
+      "step": 647
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.836466476706857e-05,
+      "loss": 1.8782,
+      "step": 648
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.8358492180304665e-05,
+      "loss": 1.8733,
+      "step": 649
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.8352308464707026e-05,
+      "loss": 1.6831,
+      "step": 650
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.834611362402419e-05,
+      "loss": 1.5292,
+      "step": 651
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.833990766201145e-05,
+      "loss": 1.5917,
+      "step": 652
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.833369058243081e-05,
+      "loss": 1.7071,
+      "step": 653
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.832746238905105e-05,
+      "loss": 2.1786,
+      "step": 654
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.832122308564768e-05,
+      "loss": 1.4945,
+      "step": 655
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.831497267600291e-05,
+      "loss": 1.4296,
+      "step": 656
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.830871116390574e-05,
+      "loss": 1.7038,
+      "step": 657
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.830243855315185e-05,
+      "loss": 1.5754,
+      "step": 658
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.829615484754367e-05,
+      "loss": 1.6699,
+      "step": 659
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.828986005089037e-05,
+      "loss": 1.5966,
+      "step": 660
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.82835541670078e-05,
+      "loss": 1.2383,
+      "step": 661
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.827723719971858e-05,
+      "loss": 1.7292,
+      "step": 662
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.827090915285202e-05,
+      "loss": 1.6143,
+      "step": 663
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.8264570030244156e-05,
+      "loss": 1.9319,
+      "step": 664
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.825821983573773e-05,
+      "loss": 1.3055,
+      "step": 665
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.8251858573182185e-05,
+      "loss": 1.2461,
+      "step": 666
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.824548624643372e-05,
+      "loss": 1.6493,
+      "step": 667
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.82391028593552e-05,
+      "loss": 1.8985,
+      "step": 668
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.823270841581619e-05,
+      "loss": 1.547,
+      "step": 669
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.822630291969299e-05,
+      "loss": 1.5634,
+      "step": 670
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.821988637486858e-05,
+      "loss": 1.3818,
+      "step": 671
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.821345878523263e-05,
+      "loss": 2.4124,
+      "step": 672
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.820702015468152e-05,
+      "loss": 1.8191,
+      "step": 673
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.820057048711833e-05,
+      "loss": 1.2857,
+      "step": 674
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.819410978645279e-05,
+      "loss": 1.718,
+      "step": 675
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.8187638056601375e-05,
+      "loss": 1.4544,
+      "step": 676
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.818115530148721e-05,
+      "loss": 1.5801,
+      "step": 677
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.8174661525040114e-05,
+      "loss": 1.2528,
+      "step": 678
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.8168156731196575e-05,
+      "loss": 1.6208,
+      "step": 679
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.816164092389977e-05,
+      "loss": 1.7234,
+      "step": 680
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.8155114107099566e-05,
+      "loss": 1.7017,
+      "step": 681
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.8148576284752484e-05,
+      "loss": 1.4918,
+      "step": 682
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.8142027460821704e-05,
+      "loss": 1.7461,
+      "step": 683
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.8135467639277115e-05,
+      "loss": 1.6785,
+      "step": 684
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.812889682409523e-05,
+      "loss": 1.9881,
+      "step": 685
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.812231501925927e-05,
+      "loss": 1.776,
+      "step": 686
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.811572222875908e-05,
+      "loss": 1.7496,
+      "step": 687
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.810911845659119e-05,
+      "loss": 1.5252,
+      "step": 688
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.810250370675876e-05,
+      "loss": 1.5484,
+      "step": 689
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.809587798327164e-05,
+      "loss": 1.5229,
+      "step": 690
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.80892412901463e-05,
+      "loss": 1.7404,
+      "step": 691
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.808259363140588e-05,
+      "loss": 1.5856,
+      "step": 692
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.807593501108015e-05,
+      "loss": 1.4449,
+      "step": 693
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.806926543320554e-05,
+      "loss": 1.767,
+      "step": 694
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.806258490182513e-05,
+      "loss": 1.475,
+      "step": 695
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.805589342098861e-05,
+      "loss": 1.6443,
+      "step": 696
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8049190994752336e-05,
+      "loss": 1.4845,
+      "step": 697
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.804247762717928e-05,
+      "loss": 1.6948,
+      "step": 698
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.803575332233907e-05,
+      "loss": 1.5791,
+      "step": 699
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.802901808430793e-05,
+      "loss": 1.6441,
+      "step": 700
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.8022271917168736e-05,
+      "loss": 1.8164,
+      "step": 701
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.8015514825010986e-05,
+      "loss": 1.5506,
+      "step": 702
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.8008746811930804e-05,
+      "loss": 1.6028,
+      "step": 703
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8001967882030915e-05,
+      "loss": 1.5394,
+      "step": 704
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.799517803942069e-05,
+      "loss": 1.5594,
+      "step": 705
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.798837728821609e-05,
+      "loss": 1.7875,
+      "step": 706
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.7981565632539695e-05,
+      "loss": 1.6468,
+      "step": 707
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.797474307652071e-05,
+      "loss": 1.3993,
+      "step": 708
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.796790962429494e-05,
+      "loss": 1.7498,
+      "step": 709
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.796106528000478e-05,
+      "loss": 1.4711,
+      "step": 710
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.7954210047799246e-05,
+      "loss": 1.8872,
+      "step": 711
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.7947343931833955e-05,
+      "loss": 1.7481,
+      "step": 712
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.794046693627111e-05,
+      "loss": 1.7472,
+      "step": 713
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.79335790652795e-05,
+      "loss": 1.3035,
+      "step": 714
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.792668032303454e-05,
+      "loss": 1.5147,
+      "step": 715
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.791977071371821e-05,
+      "loss": 1.9153,
+      "step": 716
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.791285024151909e-05,
+      "loss": 1.7184,
+      "step": 717
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.7905918910632316e-05,
+      "loss": 1.505,
+      "step": 718
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.789897672525966e-05,
+      "loss": 2.0087,
+      "step": 719
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.789202368960941e-05,
+      "loss": 1.6121,
+      "step": 720
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.78850598078965e-05,
+      "loss": 1.3935,
+      "step": 721
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.7878085084342375e-05,
+      "loss": 1.6887,
+      "step": 722
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.787109952317509e-05,
+      "loss": 1.8852,
+      "step": 723
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.786410312862926e-05,
+      "loss": 1.9691,
+      "step": 724
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.7857095904946074e-05,
+      "loss": 1.6956,
+      "step": 725
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.785007785637327e-05,
+      "loss": 1.7076,
+      "step": 726
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.784304898716515e-05,
+      "loss": 1.7504,
+      "step": 727
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.7836009301582604e-05,
+      "loss": 1.3859,
+      "step": 728
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.7828958803893036e-05,
+      "loss": 1.9151,
+      "step": 729
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.782189749837044e-05,
+      "loss": 1.5658,
+      "step": 730
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.781482538929534e-05,
+      "loss": 1.5025,
+      "step": 731
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.780774248095482e-05,
+      "loss": 1.4346,
+      "step": 732
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.7800648777642494e-05,
+      "loss": 1.8121,
+      "step": 733
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.779354428365855e-05,
+      "loss": 1.5906,
+      "step": 734
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.778642900330969e-05,
+      "loss": 1.8746,
+      "step": 735
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.777930294090916e-05,
+      "loss": 1.8771,
+      "step": 736
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.777216610077676e-05,
+      "loss": 1.9504,
+      "step": 737
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.77650184872388e-05,
+      "loss": 1.17,
+      "step": 738
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.775786010462813e-05,
+      "loss": 1.3596,
+      "step": 739
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.7750690957284135e-05,
+      "loss": 2.0342,
+      "step": 740
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.7743511049552714e-05,
+      "loss": 1.5043,
+      "step": 741
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.7736320385786293e-05,
+      "loss": 1.574,
+      "step": 742
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.772911897034383e-05,
+      "loss": 1.4594,
+      "step": 743
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.772190680759078e-05,
+      "loss": 1.4431,
+      "step": 744
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.771468390189912e-05,
+      "loss": 1.6044,
+      "step": 745
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.770745025764735e-05,
+      "loss": 1.8424,
+      "step": 746
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.770020587922047e-05,
+      "loss": 1.5363,
+      "step": 747
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.769295077100998e-05,
+      "loss": 1.2161,
+      "step": 748
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.768568493741391e-05,
+      "loss": 1.4678,
+      "step": 749
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.767840838283677e-05,
+      "loss": 1.3263,
+      "step": 750
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.7671121111689565e-05,
+      "loss": 1.8726,
+      "step": 751
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.766382312838981e-05,
+      "loss": 1.6261,
+      "step": 752
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.765651443736151e-05,
+      "loss": 1.6494,
+      "step": 753
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.764919504303517e-05,
+      "loss": 1.4791,
+      "step": 754
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.764186494984775e-05,
+      "loss": 1.8866,
+      "step": 755
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.7634524162242736e-05,
+      "loss": 1.7851,
+      "step": 756
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.762717268467008e-05,
+      "loss": 1.4038,
+      "step": 757
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.76198105215862e-05,
+      "loss": 1.5676,
+      "step": 758
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.7612437677454024e-05,
+      "loss": 1.5421,
+      "step": 759
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.760505415674293e-05,
+      "loss": 1.9788,
+      "step": 760
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.759765996392877e-05,
+      "loss": 1.7209,
+      "step": 761
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.759025510349386e-05,
+      "loss": 1.3569,
+      "step": 762
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.758283957992702e-05,
+      "loss": 1.5329,
+      "step": 763
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.757541339772348e-05,
+      "loss": 1.7699,
+      "step": 764
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.7567976561384975e-05,
+      "loss": 1.2581,
+      "step": 765
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.7560529075419664e-05,
+      "loss": 1.3052,
+      "step": 766
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.75530709443422e-05,
+      "loss": 1.6112,
+      "step": 767
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.754560217267365e-05,
+      "loss": 1.5907,
+      "step": 768
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.7538122764941556e-05,
+      "loss": 1.7246,
+      "step": 769
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.753063272567991e-05,
+      "loss": 1.9409,
+      "step": 770
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.752313205942912e-05,
+      "loss": 2.1799,
+      "step": 771
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.751562077073607e-05,
+      "loss": 1.3285,
+      "step": 772
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.750809886415406e-05,
+      "loss": 1.4149,
+      "step": 773
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.750056634424284e-05,
+      "loss": 1.5053,
+      "step": 774
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.7493023215568595e-05,
+      "loss": 1.4196,
+      "step": 775
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.7485469482703925e-05,
+      "loss": 2.0687,
+      "step": 776
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.747790515022787e-05,
+      "loss": 1.4623,
+      "step": 777
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.747033022272589e-05,
+      "loss": 2.0351,
+      "step": 778
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.746274470478989e-05,
+      "loss": 1.7049,
+      "step": 779
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.7455148601018155e-05,
+      "loss": 1.7216,
+      "step": 780
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.7447541916015414e-05,
+      "loss": 1.5145,
+      "step": 781
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.743992465439281e-05,
+      "loss": 1.7854,
+      "step": 782
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.7432296820767885e-05,
+      "loss": 1.4157,
+      "step": 783
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.7424658419764605e-05,
+      "loss": 1.7853,
+      "step": 784
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.7417009456013335e-05,
+      "loss": 1.4789,
+      "step": 785
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.740934993415082e-05,
+      "loss": 1.5461,
+      "step": 786
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.740167985882026e-05,
+      "loss": 1.5137,
+      "step": 787
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.739399923467119e-05,
+      "loss": 1.3291,
+      "step": 788
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.738630806635958e-05,
+      "loss": 1.5714,
+      "step": 789
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.737860635854779e-05,
+      "loss": 1.7214,
+      "step": 790
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.7370894115904555e-05,
+      "loss": 1.5552,
+      "step": 791
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.736317134310499e-05,
+      "loss": 1.6455,
+      "step": 792
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.735543804483062e-05,
+      "loss": 1.9545,
+      "step": 793
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.734769422576932e-05,
+      "loss": 1.7113,
+      "step": 794
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.7339939890615365e-05,
+      "loss": 1.8085,
+      "step": 795
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.733217504406941e-05,
+      "loss": 1.4948,
+      "step": 796
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.732439969083845e-05,
+      "loss": 1.5271,
+      "step": 797
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.731661383563587e-05,
+      "loss": 1.5271,
+      "step": 798
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.730881748318143e-05,
+      "loss": 1.6194,
+      "step": 799
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.730101063820124e-05,
+      "loss": 1.4745,
+      "step": 800
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.729319330542777e-05,
+      "loss": 1.2829,
+      "step": 801
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.7285365489599854e-05,
+      "loss": 1.7011,
+      "step": 802
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.727752719546268e-05,
+      "loss": 1.5408,
+      "step": 803
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.726967842776778e-05,
+      "loss": 1.7024,
+      "step": 804
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.7261819191273045e-05,
+      "loss": 1.7152,
+      "step": 805
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.725394949074271e-05,
+      "loss": 1.7611,
+      "step": 806
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7246069330947344e-05,
+      "loss": 1.5066,
+      "step": 807
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.723817871666387e-05,
+      "loss": 1.55,
+      "step": 808
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.7230277652675544e-05,
+      "loss": 1.4892,
+      "step": 809
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.722236614377194e-05,
+      "loss": 1.6139,
+      "step": 810
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.7214444194749e-05,
+      "loss": 1.6332,
+      "step": 811
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.7206511810408964e-05,
+      "loss": 1.7434,
+      "step": 812
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.7198568995560404e-05,
+      "loss": 1.7386,
+      "step": 813
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.719061575501822e-05,
+      "loss": 1.5585,
+      "step": 814
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.7182652093603627e-05,
+      "loss": 1.5061,
+      "step": 815
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.717467801614416e-05,
+      "loss": 1.5166,
+      "step": 816
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.716669352747368e-05,
+      "loss": 1.7398,
+      "step": 817
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.7158698632432334e-05,
+      "loss": 1.5472,
+      "step": 818
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.71506933358666e-05,
+      "loss": 1.3442,
+      "step": 819
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.7142677642629236e-05,
+      "loss": 1.5632,
+      "step": 820
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.7134651557579334e-05,
+      "loss": 1.7743,
+      "step": 821
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.712661508558227e-05,
+      "loss": 1.301,
+      "step": 822
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.711856823150971e-05,
+      "loss": 1.6288,
+      "step": 823
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.7110511000239615e-05,
+      "loss": 1.9902,
+      "step": 824
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.710244339665626e-05,
+      "loss": 1.5762,
+      "step": 825
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.709436542565017e-05,
+      "loss": 1.5705,
+      "step": 826
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.708627709211818e-05,
+      "loss": 1.5489,
+      "step": 827
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.707817840096341e-05,
+      "loss": 1.6432,
+      "step": 828
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.707006935709524e-05,
+      "loss": 1.8954,
+      "step": 829
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.7061949965429335e-05,
+      "loss": 1.6541,
+      "step": 830
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.705382023088763e-05,
+      "loss": 1.5908,
+      "step": 831
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.704568015839835e-05,
+      "loss": 1.8337,
+      "step": 832
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.703752975289595e-05,
+      "loss": 1.5437,
+      "step": 833
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.702936901932118e-05,
+      "loss": 1.9686,
+      "step": 834
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.702119796262103e-05,
+      "loss": 1.637,
+      "step": 835
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.7013016587748766e-05,
+      "loss": 1.7902,
+      "step": 836
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.7004824899663896e-05,
+      "loss": 2.1006,
+      "step": 837
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.699662290333218e-05,
+      "loss": 2.2165,
+      "step": 838
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.698841060372564e-05,
+      "loss": 1.6182,
+      "step": 839
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.698018800582253e-05,
+      "loss": 1.8699,
+      "step": 840
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.697195511460734e-05,
+      "loss": 1.5591,
+      "step": 841
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.6963711935070824e-05,
+      "loss": 1.6465,
+      "step": 842
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.695545847220996e-05,
+      "loss": 1.6291,
+      "step": 843
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.694719473102796e-05,
+      "loss": 1.6495,
+      "step": 844
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.6938920716534255e-05,
+      "loss": 1.4413,
+      "step": 845
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.6930636433744525e-05,
+      "loss": 1.3062,
+      "step": 846
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.692234188768065e-05,
+      "loss": 1.4199,
+      "step": 847
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.691403708337078e-05,
+      "loss": 1.2364,
+      "step": 848
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.6905722025849204e-05,
+      "loss": 1.65,
+      "step": 849
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.68973967201565e-05,
+      "loss": 1.7933,
+      "step": 850
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.688906117133942e-05,
+      "loss": 1.5142,
+      "step": 851
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.688071538445095e-05,
+      "loss": 1.628,
+      "step": 852
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.6872359364550245e-05,
+      "loss": 1.3949,
+      "step": 853
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.68639931167027e-05,
+      "loss": 1.7068,
+      "step": 854
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.68556166459799e-05,
+      "loss": 1.9848,
+      "step": 855
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.684722995745961e-05,
+      "loss": 1.6619,
+      "step": 856
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.683883305622582e-05,
+      "loss": 1.4843,
+      "step": 857
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.683042594736867e-05,
+      "loss": 1.9497,
+      "step": 858
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.682200863598453e-05,
+      "loss": 1.6036,
+      "step": 859
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.6813581127175916e-05,
+      "loss": 1.685,
+      "step": 860
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.6805143426051564e-05,
+      "loss": 1.5563,
+      "step": 861
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.6796695537726364e-05,
+      "loss": 1.8723,
+      "step": 862
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.678823746732139e-05,
+      "loss": 1.4578,
+      "step": 863
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.6779769219963884e-05,
+      "loss": 1.7557,
+      "step": 864
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.677129080078725e-05,
+      "loss": 1.6742,
+      "step": 865
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.676280221493107e-05,
+      "loss": 1.5015,
+      "step": 866
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.6754303467541095e-05,
+      "loss": 1.7286,
+      "step": 867
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.674579456376922e-05,
+      "loss": 1.7102,
+      "step": 868
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.673727550877351e-05,
+      "loss": 1.6313,
+      "step": 869
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.6728746307718164e-05,
+      "loss": 1.491,
+      "step": 870
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.6720206965773554e-05,
+      "loss": 1.6063,
+      "step": 871
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.6711657488116185e-05,
+      "loss": 1.764,
+      "step": 872
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.6703097879928715e-05,
+      "loss": 1.6052,
+      "step": 873
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.669452814639994e-05,
+      "loss": 1.5609,
+      "step": 874
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.668594829272479e-05,
+      "loss": 1.3898,
+      "step": 875
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.667735832410433e-05,
+      "loss": 1.9448,
+      "step": 876
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.666875824574577e-05,
+      "loss": 1.3524,
+      "step": 877
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.666014806286243e-05,
+      "loss": 1.6109,
+      "step": 878
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.6651527780673764e-05,
+      "loss": 1.3357,
+      "step": 879
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.664289740440535e-05,
+      "loss": 1.5391,
+      "step": 880
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.6634256939288885e-05,
+      "loss": 1.7805,
+      "step": 881
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.662560639056217e-05,
+      "loss": 1.6405,
+      "step": 882
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.661694576346914e-05,
+      "loss": 1.7683,
+      "step": 883
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.660827506325981e-05,
+      "loss": 1.4612,
+      "step": 884
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.659959429519033e-05,
+      "loss": 1.5311,
+      "step": 885
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.659090346452295e-05,
+      "loss": 1.6764,
+      "step": 886
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.6582202576526e-05,
+      "loss": 1.5068,
+      "step": 887
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.657349163647392e-05,
+      "loss": 1.5189,
+      "step": 888
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.6564770649647244e-05,
+      "loss": 1.4261,
+      "step": 889
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.655603962133259e-05,
+      "loss": 1.5871,
+      "step": 890
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.654729855682267e-05,
+      "loss": 1.6222,
+      "step": 891
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.653854746141629e-05,
+      "loss": 1.3151,
+      "step": 892
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.65297863404183e-05,
+      "loss": 1.65,
+      "step": 893
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.652101519913967e-05,
+      "loss": 1.5861,
+      "step": 894
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.651223404289741e-05,
+      "loss": 1.5755,
+      "step": 895
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.6503442877014636e-05,
+      "loss": 1.6442,
+      "step": 896
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.64946417068205e-05,
+      "loss": 1.5536,
+      "step": 897
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.648583053765023e-05,
+      "loss": 1.4257,
+      "step": 898
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.647700937484512e-05,
+      "loss": 1.5146,
+      "step": 899
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.646817822375251e-05,
+      "loss": 1.4811,
+      "step": 900
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.645933708972583e-05,
+      "loss": 1.8025,
+      "step": 901
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.6450485978124494e-05,
+      "loss": 1.8122,
+      "step": 902
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.644162489431404e-05,
+      "loss": 2.089,
+      "step": 903
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.6432753843666004e-05,
+      "loss": 1.8404,
+      "step": 904
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.642387283155797e-05,
+      "loss": 1.4859,
+      "step": 905
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.641498186337357e-05,
+      "loss": 1.6153,
+      "step": 906
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.640608094450248e-05,
+      "loss": 1.8293,
+      "step": 907
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.639717008034038e-05,
+      "loss": 1.4503,
+      "step": 908
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.638824927628901e-05,
+      "loss": 1.5915,
+      "step": 909
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.6379318537756094e-05,
+      "loss": 1.4079,
+      "step": 910
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.637037787015544e-05,
+      "loss": 1.4972,
+      "step": 911
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.6361427278906815e-05,
+      "loss": 2.0312,
+      "step": 912
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.6352466769436044e-05,
+      "loss": 1.7953,
+      "step": 913
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.634349634717492e-05,
+      "loss": 1.6306,
+      "step": 914
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.63345160175613e-05,
+      "loss": 1.5734,
+      "step": 915
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.6325525786039004e-05,
+      "loss": 1.2524,
+      "step": 916
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.6316525658057876e-05,
+      "loss": 1.6502,
+      "step": 917
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.630751563907375e-05,
+      "loss": 1.641,
+      "step": 918
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.629849573454845e-05,
+      "loss": 1.3019,
+      "step": 919
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.6289465949949805e-05,
+      "loss": 1.5091,
+      "step": 920
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.628042629075164e-05,
+      "loss": 1.5019,
+      "step": 921
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.627137676243374e-05,
+      "loss": 1.5229,
+      "step": 922
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.626231737048189e-05,
+      "loss": 1.226,
+      "step": 923
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.625324812038786e-05,
+      "loss": 1.6996,
+      "step": 924
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.6244169017649385e-05,
+      "loss": 1.6795,
+      "step": 925
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.623508006777018e-05,
+      "loss": 1.6949,
+      "step": 926
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.622598127625991e-05,
+      "loss": 1.8036,
+      "step": 927
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.6216872648634235e-05,
+      "loss": 1.9046,
+      "step": 928
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.620775419041476e-05,
+      "loss": 1.4679,
+      "step": 929
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.619862590712904e-05,
+      "loss": 1.2019,
+      "step": 930
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.618948780431061e-05,
+      "loss": 1.1759,
+      "step": 931
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.6180339887498953e-05,
+      "loss": 1.7787,
+      "step": 932
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.617118216223948e-05,
+      "loss": 1.5699,
+      "step": 933
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.616201463408357e-05,
+      "loss": 1.752,
+      "step": 934
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.6152837308588535e-05,
+      "loss": 1.7895,
+      "step": 935
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.6143650191317625e-05,
+      "loss": 1.8609,
+      "step": 936
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.613445328784003e-05,
+      "loss": 1.6286,
+      "step": 937
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.6125246603730875e-05,
+      "loss": 1.6711,
+      "step": 938
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.61160301445712e-05,
+      "loss": 1.7357,
+      "step": 939
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.610680391594799e-05,
+      "loss": 1.7114,
+      "step": 940
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.609756792345414e-05,
+      "loss": 1.6556,
+      "step": 941
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.6088322172688454e-05,
+      "loss": 1.361,
+      "step": 942
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.6079066669255685e-05,
+      "loss": 1.7143,
+      "step": 943
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.606980141876647e-05,
+      "loss": 1.5539,
+      "step": 944
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.606052642683735e-05,
+      "loss": 1.4883,
+      "step": 945
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.605124169909079e-05,
+      "loss": 1.5267,
+      "step": 946
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.604194724115515e-05,
+      "loss": 1.5666,
+      "step": 947
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.6032643058664684e-05,
+      "loss": 1.9905,
+      "step": 948
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.602332915725955e-05,
+      "loss": 1.5381,
+      "step": 949
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.60140055425858e-05,
+      "loss": 1.7322,
+      "step": 950
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.600467222029535e-05,
+      "loss": 1.3818,
+      "step": 951
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.5995329196046016e-05,
+      "loss": 1.5387,
+      "step": 952
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.598597647550151e-05,
+      "loss": 1.8122,
+      "step": 953
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.59766140643314e-05,
+      "loss": 1.5953,
+      "step": 954
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.596724196821114e-05,
+      "loss": 1.8209,
+      "step": 955
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.595786019282206e-05,
+      "loss": 1.7275,
+      "step": 956
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.594846874385133e-05,
+      "loss": 1.5624,
+      "step": 957
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.593906762699202e-05,
+      "loss": 1.4297,
+      "step": 958
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.5929656847943036e-05,
+      "loss": 1.6746,
+      "step": 959
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.592023641240915e-05,
+      "loss": 1.4554,
+      "step": 960
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.591080632610099e-05,
+      "loss": 1.3771,
+      "step": 961
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.590136659473502e-05,
+      "loss": 2.0115,
+      "step": 962
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.5891917224033575e-05,
+      "loss": 2.0063,
+      "step": 963
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.588245821972481e-05,
+      "loss": 1.3671,
+      "step": 964
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.587298958754274e-05,
+      "loss": 1.5355,
+      "step": 965
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.5863511333227196e-05,
+      "loss": 1.8621,
+      "step": 966
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.5854023462523875e-05,
+      "loss": 1.4456,
+      "step": 967
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.5844525981184244e-05,
+      "loss": 1.8278,
+      "step": 968
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.583501889496567e-05,
+      "loss": 1.4893,
+      "step": 969
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.582550220963127e-05,
+      "loss": 1.5093,
+      "step": 970
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.581597593095004e-05,
+      "loss": 1.8708,
+      "step": 971
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.580644006469676e-05,
+      "loss": 1.7599,
+      "step": 972
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.579689461665203e-05,
+      "loss": 1.6053,
+      "step": 973
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.578733959260224e-05,
+      "loss": 1.9628,
+      "step": 974
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.577777499833963e-05,
+      "loss": 1.4568,
+      "step": 975
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.576820083966219e-05,
+      "loss": 1.9227,
+      "step": 976
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.575861712237375e-05,
+      "loss": 1.4436,
+      "step": 977
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.57490238522839e-05,
+      "loss": 1.5764,
+      "step": 978
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.573942103520803e-05,
+      "loss": 1.5892,
+      "step": 979
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.572980867696735e-05,
+      "loss": 1.6236,
+      "step": 980
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.5720186783388815e-05,
+      "loss": 1.4467,
+      "step": 981
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.571055536030516e-05,
+      "loss": 1.3582,
+      "step": 982
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.570091441355493e-05,
+      "loss": 1.4584,
+      "step": 983
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.56912639489824e-05,
+      "loss": 1.7837,
+      "step": 984
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.568160397243767e-05,
+      "loss": 1.6467,
+      "step": 985
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.567193448977653e-05,
+      "loss": 1.4726,
+      "step": 986
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.5662255506860615e-05,
+      "loss": 1.4776,
+      "step": 987
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.565256702955726e-05,
+      "loss": 1.4576,
+      "step": 988
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.564286906373959e-05,
+      "loss": 1.1976,
+      "step": 989
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.5633161615286443e-05,
+      "loss": 1.6825,
+      "step": 990
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.562344469008245e-05,
+      "loss": 1.5231,
+      "step": 991
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.561371829401796e-05,
+      "loss": 1.8706,
+      "step": 992
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.560398243298907e-05,
+      "loss": 1.7967,
+      "step": 993
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.559423711289761e-05,
+      "loss": 1.4289,
+      "step": 994
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.558448233965116e-05,
+      "loss": 1.4309,
+      "step": 995
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.5574718119163013e-05,
+      "loss": 1.3758,
+      "step": 996
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.556494445735218e-05,
+      "loss": 1.9035,
+      "step": 997
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.555516136014344e-05,
+      "loss": 1.3978,
+      "step": 998
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.554536883346724e-05,
+      "loss": 1.3376,
+      "step": 999
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.5535566883259767e-05,
+      "loss": 1.2992,
+      "step": 1000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.5525755515462925e-05,
+      "loss": 1.3485,
+      "step": 1001
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.5515934736024315e-05,
+      "loss": 1.9009,
+      "step": 1002
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.5506104550897245e-05,
+      "loss": 1.6154,
+      "step": 1003
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.549626496604073e-05,
+      "loss": 1.7949,
+      "step": 1004
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.54864159874195e-05,
+      "loss": 1.4455,
+      "step": 1005
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.5476557621003934e-05,
+      "loss": 1.592,
+      "step": 1006
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.546668987277014e-05,
+      "loss": 1.3845,
+      "step": 1007
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.54568127486999e-05,
+      "loss": 1.9232,
+      "step": 1008
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.5446926254780686e-05,
+      "loss": 1.9834,
+      "step": 1009
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.543703039700563e-05,
+      "loss": 1.8988,
+      "step": 1010
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.542712518137358e-05,
+      "loss": 1.5278,
+      "step": 1011
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.541721061388901e-05,
+      "loss": 1.5744,
+      "step": 1012
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.54072867005621e-05,
+      "loss": 1.5297,
+      "step": 1013
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.539735344740868e-05,
+      "loss": 1.638,
+      "step": 1014
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.538741086045023e-05,
+      "loss": 1.2058,
+      "step": 1015
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.537745894571391e-05,
+      "loss": 2.0204,
+      "step": 1016
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.536749770923252e-05,
+      "loss": 1.7233,
+      "step": 1017
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.535752715704452e-05,
+      "loss": 1.5694,
+      "step": 1018
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.534754729519402e-05,
+      "loss": 1.3117,
+      "step": 1019
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.533755812973075e-05,
+      "loss": 1.7292,
+      "step": 1020
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.532755966671011e-05,
+      "loss": 1.2251,
+      "step": 1021
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.531755191219312e-05,
+      "loss": 1.3994,
+      "step": 1022
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.530753487224643e-05,
+      "loss": 1.7878,
+      "step": 1023
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.529750855294232e-05,
+      "loss": 1.4629,
+      "step": 1024
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.5287472960358706e-05,
+      "loss": 1.5845,
+      "step": 1025
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.5277428100579123e-05,
+      "loss": 1.7185,
+      "step": 1026
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3.5267373979692706e-05,
+      "loss": 1.1366,
+      "step": 1027
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.5257310603794236e-05,
+      "loss": 1.9161,
+      "step": 1028
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3.524723797898406e-05,
+      "loss": 1.3159,
+      "step": 1029
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3.523715611136818e-05,
+      "loss": 1.8724,
+      "step": 1030
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.522706500705816e-05,
+      "loss": 1.6522,
+      "step": 1031
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.5216964672171195e-05,
+      "loss": 1.4032,
+      "step": 1032
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.520685511283005e-05,
+      "loss": 1.1783,
+      "step": 1033
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.519673633516311e-05,
+      "loss": 1.8208,
+      "step": 1034
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.518660834530431e-05,
+      "loss": 1.8209,
+      "step": 1035
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.5176471149393205e-05,
+      "loss": 1.4356,
+      "step": 1036
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.516632475357491e-05,
+      "loss": 1.538,
+      "step": 1037
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.5156169164000137e-05,
+      "loss": 1.491,
+      "step": 1038
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.514600438682514e-05,
+      "loss": 1.8408,
+      "step": 1039
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.5135830428211766e-05,
+      "loss": 1.4597,
+      "step": 1040
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.512564729432743e-05,
+      "loss": 1.4548,
+      "step": 1041
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.5115454991345095e-05,
+      "loss": 1.8536,
+      "step": 1042
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.5105253525443284e-05,
+      "loss": 1.2502,
+      "step": 1043
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.5095042902806086e-05,
+      "loss": 1.4747,
+      "step": 1044
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.508482312962313e-05,
+      "loss": 1.8897,
+      "step": 1045
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.5074594212089604e-05,
+      "loss": 1.8012,
+      "step": 1046
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.5064356156406215e-05,
+      "loss": 1.5306,
+      "step": 1047
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.505410896877924e-05,
+      "loss": 1.8328,
+      "step": 1048
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.504385265542047e-05,
+      "loss": 1.5018,
+      "step": 1049
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.503358722254723e-05,
+      "loss": 1.4653,
+      "step": 1050
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.5023312676382385e-05,
+      "loss": 1.4997,
+      "step": 1051
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.501302902315432e-05,
+      "loss": 1.5192,
+      "step": 1052
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.5002736269096934e-05,
+      "loss": 1.5747,
+      "step": 1053
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.499243442044965e-05,
+      "loss": 1.8655,
+      "step": 1054
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.4982123483457405e-05,
+      "loss": 1.7686,
+      "step": 1055
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.497180346437063e-05,
+      "loss": 1.5205,
+      "step": 1056
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.496147436944529e-05,
+      "loss": 1.5677,
+      "step": 1057
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.4951136204942817e-05,
+      "loss": 1.5271,
+      "step": 1058
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.4940788977130164e-05,
+      "loss": 1.4054,
+      "step": 1059
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.493043269227978e-05,
+      "loss": 1.6759,
+      "step": 1060
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.49200673566696e-05,
+      "loss": 1.5794,
+      "step": 1061
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.4909692976583025e-05,
+      "loss": 1.8507,
+      "step": 1062
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.489930955830898e-05,
+      "loss": 1.6998,
+      "step": 1063
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.4888917108141825e-05,
+      "loss": 1.8435,
+      "step": 1064
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.4878515632381425e-05,
+      "loss": 1.7628,
+      "step": 1065
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.4868105137333104e-05,
+      "loss": 1.6635,
+      "step": 1066
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.4857685629307664e-05,
+      "loss": 1.74,
+      "step": 1067
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.4847257114621356e-05,
+      "loss": 1.7783,
+      "step": 1068
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.48368195995959e-05,
+      "loss": 1.7297,
+      "step": 1069
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.482637309055846e-05,
+      "loss": 1.5303,
+      "step": 1070
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.481591759384168e-05,
+      "loss": 1.7376,
+      "step": 1071
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.4805453115783626e-05,
+      "loss": 1.2696,
+      "step": 1072
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.4794979662727804e-05,
+      "loss": 2.0244,
+      "step": 1073
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.4784497241023195e-05,
+      "loss": 1.8679,
+      "step": 1074
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.4774005857024185e-05,
+      "loss": 1.5034,
+      "step": 1075
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.4763505517090596e-05,
+      "loss": 1.3648,
+      "step": 1076
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.47529962275877e-05,
+      "loss": 1.4428,
+      "step": 1077
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.4742477994886165e-05,
+      "loss": 1.6475,
+      "step": 1078
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.4731950825362106e-05,
+      "loss": 1.4628,
+      "step": 1079
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.472141472539704e-05,
+      "loss": 1.4779,
+      "step": 1080
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.47108697013779e-05,
+      "loss": 1.6217,
+      "step": 1081
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.4700315759697045e-05,
+      "loss": 1.6336,
+      "step": 1082
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.468975290675221e-05,
+      "loss": 1.5702,
+      "step": 1083
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.467918114894656e-05,
+      "loss": 1.2273,
+      "step": 1084
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.4668600492688635e-05,
+      "loss": 1.9606,
+      "step": 1085
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.4658010944392385e-05,
+      "loss": 1.744,
+      "step": 1086
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.464741251047715e-05,
+      "loss": 1.5932,
+      "step": 1087
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.463680519736764e-05,
+      "loss": 1.4873,
+      "step": 1088
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.462618901149398e-05,
+      "loss": 1.825,
+      "step": 1089
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.461556395929163e-05,
+      "loss": 1.8057,
+      "step": 1090
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.460493004720147e-05,
+      "loss": 1.4646,
+      "step": 1091
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.4594287281669713e-05,
+      "loss": 1.4662,
+      "step": 1092
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.458363566914796e-05,
+      "loss": 1.7328,
+      "step": 1093
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.457297521609316e-05,
+      "loss": 2.0391,
+      "step": 1094
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.456230592896765e-05,
+      "loss": 1.8489,
+      "step": 1095
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.455162781423909e-05,
+      "loss": 1.8099,
+      "step": 1096
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.454094087838051e-05,
+      "loss": 1.704,
+      "step": 1097
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.453024512787027e-05,
+      "loss": 1.5132,
+      "step": 1098
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.4519540569192097e-05,
+      "loss": 1.381,
+      "step": 1099
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.450882720883504e-05,
+      "loss": 1.5345,
+      "step": 1100
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.44981050532935e-05,
+      "loss": 1.7061,
+      "step": 1101
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.4487374109067174e-05,
+      "loss": 1.3377,
+      "step": 1102
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.4476634382661135e-05,
+      "loss": 1.478,
+      "step": 1103
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.446588588058574e-05,
+      "loss": 1.636,
+      "step": 1104
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.445512860935669e-05,
+      "loss": 1.353,
+      "step": 1105
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.4444362575495e-05,
+      "loss": 1.6643,
+      "step": 1106
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.443358778552698e-05,
+      "loss": 1.5645,
+      "step": 1107
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.442280424598426e-05,
+      "loss": 1.5542,
+      "step": 1108
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.4412011963403773e-05,
+      "loss": 1.7813,
+      "step": 1109
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.4401210944327745e-05,
+      "loss": 1.3801,
+      "step": 1110
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.439040119530372e-05,
+      "loss": 1.6005,
+      "step": 1111
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.4379582722884496e-05,
+      "loss": 1.8725,
+      "step": 1112
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.43687555336282e-05,
+      "loss": 1.7498,
+      "step": 1113
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.435791963409822e-05,
+      "loss": 1.7959,
+      "step": 1114
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.4347075030863224e-05,
+      "loss": 1.7125,
+      "step": 1115
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.433622173049717e-05,
+      "loss": 1.472,
+      "step": 1116
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.4325359739579265e-05,
+      "loss": 1.5936,
+      "step": 1117
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.4314489064694006e-05,
+      "loss": 1.4108,
+      "step": 1118
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.430360971243115e-05,
+      "loss": 1.7437,
+      "step": 1119
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.4292721689385705e-05,
+      "loss": 1.2854,
+      "step": 1120
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.428182500215794e-05,
+      "loss": 1.2375,
+      "step": 1121
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.427091965735338e-05,
+      "loss": 1.255,
+      "step": 1122
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.4260005661582796e-05,
+      "loss": 1.5184,
+      "step": 1123
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.42490830214622e-05,
+      "loss": 1.4044,
+      "step": 1124
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.423815174361284e-05,
+      "loss": 1.2383,
+      "step": 1125
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.422721183466122e-05,
+      "loss": 1.3763,
+      "step": 1126
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.4216263301239047e-05,
+      "loss": 1.4942,
+      "step": 1127
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.420530614998328e-05,
+      "loss": 1.3555,
+      "step": 1128
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.41943403875361e-05,
+      "loss": 2.0496,
+      "step": 1129
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.4183366020544895e-05,
+      "loss": 1.4818,
+      "step": 1130
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.417238305566228e-05,
+      "loss": 1.889,
+      "step": 1131
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.4161391499546066e-05,
+      "loss": 1.8073,
+      "step": 1132
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.41503913588593e-05,
+      "loss": 1.7136,
+      "step": 1133
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.41393826402702e-05,
+      "loss": 1.2604,
+      "step": 1134
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.412836535045222e-05,
+      "loss": 1.5918,
+      "step": 1135
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.4117339496083974e-05,
+      "loss": 1.8913,
+      "step": 1136
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.410630508384929e-05,
+      "loss": 1.7122,
+      "step": 1137
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.4095262120437183e-05,
+      "loss": 2.079,
+      "step": 1138
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.408421061254184e-05,
+      "loss": 1.4886,
+      "step": 1139
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.407315056686264e-05,
+      "loss": 1.7966,
+      "step": 1140
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.4062081990104125e-05,
+      "loss": 1.4015,
+      "step": 1141
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.405100488897603e-05,
+      "loss": 1.7746,
+      "step": 1142
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.4039919270193215e-05,
+      "loss": 1.7026,
+      "step": 1143
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.402882514047576e-05,
+      "loss": 1.2251,
+      "step": 1144
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.401772250654886e-05,
+      "loss": 1.4055,
+      "step": 1145
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.4006611375142894e-05,
+      "loss": 1.4428,
+      "step": 1146
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.3995491752993356e-05,
+      "loss": 1.4167,
+      "step": 1147
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.398436364684093e-05,
+      "loss": 1.4982,
+      "step": 1148
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.397322706343141e-05,
+      "loss": 1.3595,
+      "step": 1149
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.396208200951576e-05,
+      "loss": 1.6562,
+      "step": 1150
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.395092849185005e-05,
+      "loss": 1.5361,
+      "step": 1151
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.393976651719548e-05,
+      "loss": 1.5318,
+      "step": 1152
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.39285960923184e-05,
+      "loss": 1.8145,
+      "step": 1153
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.3917417223990275e-05,
+      "loss": 1.8544,
+      "step": 1154
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.390622991898767e-05,
+      "loss": 1.6446,
+      "step": 1155
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.389503418409229e-05,
+      "loss": 1.3861,
+      "step": 1156
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.388383002609093e-05,
+      "loss": 1.7206,
+      "step": 1157
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.3872617451775497e-05,
+      "loss": 1.7665,
+      "step": 1158
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.386139646794302e-05,
+      "loss": 1.9407,
+      "step": 1159
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.385016708139558e-05,
+      "loss": 1.9355,
+      "step": 1160
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.383892929894041e-05,
+      "loss": 1.6728,
+      "step": 1161
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.382768312738978e-05,
+      "loss": 1.6918,
+      "step": 1162
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.3816428573561076e-05,
+      "loss": 1.7918,
+      "step": 1163
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.380516564427675e-05,
+      "loss": 2.0006,
+      "step": 1164
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.3793894346364354e-05,
+      "loss": 1.908,
+      "step": 1165
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.3782614686656484e-05,
+      "loss": 1.1569,
+      "step": 1166
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.3771326671990815e-05,
+      "loss": 1.4812,
+      "step": 1167
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.376003030921009e-05,
+      "loss": 1.8962,
+      "step": 1168
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.374872560516213e-05,
+      "loss": 1.7391,
+      "step": 1169
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.3737412566699776e-05,
+      "loss": 1.7139,
+      "step": 1170
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.372609120068095e-05,
+      "loss": 1.8667,
+      "step": 1171
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.371476151396861e-05,
+      "loss": 1.337,
+      "step": 1172
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.370342351343075e-05,
+      "loss": 1.53,
+      "step": 1173
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.3692077205940434e-05,
+      "loss": 1.0921,
+      "step": 1174
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.368072259837572e-05,
+      "loss": 1.9563,
+      "step": 1175
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.366935969761973e-05,
+      "loss": 1.6369,
+      "step": 1176
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.365798851056061e-05,
+      "loss": 1.5583,
+      "step": 1177
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.364660904409151e-05,
+      "loss": 1.1697,
+      "step": 1178
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.3635221305110616e-05,
+      "loss": 1.7389,
+      "step": 1179
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.362382530052111e-05,
+      "loss": 1.5185,
+      "step": 1180
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.361242103723122e-05,
+      "loss": 1.3738,
+      "step": 1181
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.360100852215413e-05,
+      "loss": 1.6497,
+      "step": 1182
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.3589587762208085e-05,
+      "loss": 1.5889,
+      "step": 1183
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.357815876431628e-05,
+      "loss": 1.7571,
+      "step": 1184
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.356672153540692e-05,
+      "loss": 1.4027,
+      "step": 1185
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.355527608241321e-05,
+      "loss": 1.2755,
+      "step": 1186
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.354382241227332e-05,
+      "loss": 1.3832,
+      "step": 1187
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.3532360531930424e-05,
+      "loss": 1.5005,
+      "step": 1188
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.352089044833265e-05,
+      "loss": 1.7017,
+      "step": 1189
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.350941216843311e-05,
+      "loss": 1.7426,
+      "step": 1190
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.3497925699189904e-05,
+      "loss": 1.8196,
+      "step": 1191
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.348643104756605e-05,
+      "loss": 1.7152,
+      "step": 1192
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.347492822052957e-05,
+      "loss": 1.5313,
+      "step": 1193
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.346341722505343e-05,
+      "loss": 1.7087,
+      "step": 1194
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.345189806811552e-05,
+      "loss": 2.1994,
+      "step": 1195
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.3440370756698724e-05,
+      "loss": 1.8804,
+      "step": 1196
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.342883529779083e-05,
+      "loss": 1.5462,
+      "step": 1197
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.3417291698384586e-05,
+      "loss": 1.5166,
+      "step": 1198
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.3405739965477675e-05,
+      "loss": 1.2847,
+      "step": 1199
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.339418010607269e-05,
+      "loss": 1.3554,
+      "step": 1200
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.3382612127177166e-05,
+      "loss": 2.0597,
+      "step": 1201
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.3371036035803576e-05,
+      "loss": 1.8137,
+      "step": 1202
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.335945183896928e-05,
+      "loss": 1.4125,
+      "step": 1203
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.334785954369656e-05,
+      "loss": 1.7243,
+      "step": 1204
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.333625915701263e-05,
+      "loss": 1.2448,
+      "step": 1205
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.3324650685949574e-05,
+      "loss": 1.5757,
+      "step": 1206
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.3313034137544406e-05,
+      "loss": 1.5452,
+      "step": 1207
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.330140951883902e-05,
+      "loss": 1.4812,
+      "step": 1208
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.32897768368802e-05,
+      "loss": 1.3136,
+      "step": 1209
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.327813609871964e-05,
+      "loss": 1.7066,
+      "step": 1210
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.326648731141388e-05,
+      "loss": 1.4514,
+      "step": 1211
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.325483048202439e-05,
+      "loss": 1.415,
+      "step": 1212
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.324316561761745e-05,
+      "loss": 1.6863,
+      "step": 1213
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.3231492725264284e-05,
+      "loss": 2.1379,
+      "step": 1214
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.321981181204092e-05,
+      "loss": 1.4011,
+      "step": 1215
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.320812288502829e-05,
+      "loss": 1.6989,
+      "step": 1216
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.319642595131216e-05,
+      "loss": 1.539,
+      "step": 1217
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.3184721017983145e-05,
+      "loss": 1.1753,
+      "step": 1218
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.317300809213675e-05,
+      "loss": 1.4356,
+      "step": 1219
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.316128718087328e-05,
+      "loss": 1.7149,
+      "step": 1220
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.31495582912979e-05,
+      "loss": 1.7525,
+      "step": 1221
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.313782143052061e-05,
+      "loss": 1.4002,
+      "step": 1222
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.312607660565625e-05,
+      "loss": 1.9818,
+      "step": 1223
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.311432382382445e-05,
+      "loss": 1.5239,
+      "step": 1224
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.310256309214973e-05,
+      "loss": 1.7048,
+      "step": 1225
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.309079441776136e-05,
+      "loss": 1.917,
+      "step": 1226
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.3079017807793475e-05,
+      "loss": 1.7472,
+      "step": 1227
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.3067233269384996e-05,
+      "loss": 1.3352,
+      "step": 1228
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.305544080967966e-05,
+      "loss": 1.6275,
+      "step": 1229
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.3043640435825996e-05,
+      "loss": 1.6427,
+      "step": 1230
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.3031832154977334e-05,
+      "loss": 1.5543,
+      "step": 1231
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.3020015974291814e-05,
+      "loss": 1.1581,
+      "step": 1232
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.300819190093233e-05,
+      "loss": 1.5952,
+      "step": 1233
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.299635994206661e-05,
+      "loss": 1.2583,
+      "step": 1234
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.298452010486711e-05,
+      "loss": 1.5792,
+      "step": 1235
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.2972672396511085e-05,
+      "loss": 2.0085,
+      "step": 1236
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.296081682418057e-05,
+      "loss": 1.4235,
+      "step": 1237
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.294895339506235e-05,
+      "loss": 1.5856,
+      "step": 1238
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.2937082116348e-05,
+      "loss": 1.4722,
+      "step": 1239
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.292520299523381e-05,
+      "loss": 1.9405,
+      "step": 1240
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.2913316038920865e-05,
+      "loss": 1.0868,
+      "step": 1241
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.290142125461498e-05,
+      "loss": 1.363,
+      "step": 1242
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.288951864952672e-05,
+      "loss": 1.5512,
+      "step": 1243
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.287760823087138e-05,
+      "loss": 1.5917,
+      "step": 1244
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.2865690005869015e-05,
+      "loss": 1.3541,
+      "step": 1245
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.285376398174439e-05,
+      "loss": 1.7731,
+      "step": 1246
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.284183016572701e-05,
+      "loss": 1.3836,
+      "step": 1247
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.282988856505109e-05,
+      "loss": 1.7346,
+      "step": 1248
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.2817939186955584e-05,
+      "loss": 1.4571,
+      "step": 1249
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.2805982038684145e-05,
+      "loss": 1.2979,
+      "step": 1250
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.2794017127485145e-05,
+      "loss": 1.382,
+      "step": 1251
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.2782044460611664e-05,
+      "loss": 1.4615,
+      "step": 1252
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.277006404532147e-05,
+      "loss": 1.875,
+      "step": 1253
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.275807588887703e-05,
+      "loss": 1.6321,
+      "step": 1254
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.274607999854551e-05,
+      "loss": 1.2425,
+      "step": 1255
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.273407638159878e-05,
+      "loss": 1.3315,
+      "step": 1256
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.2722065045313356e-05,
+      "loss": 1.4676,
+      "step": 1257
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.271004599697047e-05,
+      "loss": 1.3775,
+      "step": 1258
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 3.269801924385602e-05,
+      "loss": 1.1929,
+      "step": 1259
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 3.268598479326055e-05,
+      "loss": 1.6113,
+      "step": 1260
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 3.26739426524793e-05,
+      "loss": 1.6078,
+      "step": 1261
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 3.2661892828812155e-05,
+      "loss": 1.6536,
+      "step": 1262
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 3.264983532956366e-05,
+      "loss": 1.3076,
+      "step": 1263
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 3.263777016204303e-05,
+      "loss": 1.8086,
+      "step": 1264
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 3.26256973335641e-05,
+      "loss": 1.6858,
+      "step": 1265
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 3.261361685144538e-05,
+      "loss": 1.435,
+      "step": 1266
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 3.2601528723009976e-05,
+      "loss": 1.6076,
+      "step": 1267
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 3.258943295558568e-05,
+      "loss": 1.7101,
+      "step": 1268
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 3.257732955650488e-05,
+      "loss": 1.6817,
+      "step": 1269
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 3.25652185331046e-05,
+      "loss": 1.7112,
+      "step": 1270
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 3.2553099892726494e-05,
+      "loss": 1.3696,
+      "step": 1271
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 3.254097364271682e-05,
+      "loss": 1.6321,
+      "step": 1272
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 3.252883979042646e-05,
+      "loss": 1.6595,
+      "step": 1273
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 3.25166983432109e-05,
+      "loss": 1.6624,
+      "step": 1274
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 3.250454930843022e-05,
+      "loss": 1.3023,
+      "step": 1275
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 3.249239269344912e-05,
+      "loss": 1.81,
+      "step": 1276
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 3.248022850563688e-05,
+      "loss": 1.9634,
+      "step": 1277
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 3.246805675236737e-05,
+      "loss": 1.5737,
+      "step": 1278
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 3.245587744101905e-05,
+      "loss": 1.7754,
+      "step": 1279
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 3.244369057897499e-05,
+      "loss": 1.484,
+      "step": 1280
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 3.243149617362277e-05,
+      "loss": 1.3484,
+      "step": 1281
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 3.241929423235461e-05,
+      "loss": 2.007,
+      "step": 1282
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 3.240708476256725e-05,
+      "loss": 1.6539,
+      "step": 1283
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 3.239486777166204e-05,
+      "loss": 1.7815,
+      "step": 1284
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 3.2382643267044845e-05,
+      "loss": 1.5745,
+      "step": 1285
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 3.23704112561261e-05,
+      "loss": 1.537,
+      "step": 1286
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 3.2358171746320817e-05,
+      "loss": 1.6173,
+      "step": 1287
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 3.234592474504851e-05,
+      "loss": 1.5762,
+      "step": 1288
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 3.233367025973326e-05,
+      "loss": 1.5684,
+      "step": 1289
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 3.232140829780367e-05,
+      "loss": 1.7439,
+      "step": 1290
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 3.230913886669292e-05,
+      "loss": 1.6642,
+      "step": 1291
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 3.2296861973838644e-05,
+      "loss": 1.4548,
+      "step": 1292
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 3.2284577626683055e-05,
+      "loss": 1.4642,
+      "step": 1293
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 3.227228583267287e-05,
+      "loss": 1.5412,
+      "step": 1294
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 3.2259986599259304e-05,
+      "loss": 1.597,
+      "step": 1295
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 3.2247679933898116e-05,
+      "loss": 1.4669,
+      "step": 1296
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 3.223536584404954e-05,
+      "loss": 1.6995,
+      "step": 1297
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 3.222304433717832e-05,
+      "loss": 1.4416,
+      "step": 1298
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 3.221071542075369e-05,
+      "loss": 1.4211,
+      "step": 1299
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 3.21983791022494e-05,
+      "loss": 1.8437,
+      "step": 1300
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 3.2186035389143655e-05,
+      "loss": 1.3254,
+      "step": 1301
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 3.217368428891917e-05,
+      "loss": 1.8026,
+      "step": 1302
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 3.216132580906311e-05,
+      "loss": 1.2876,
+      "step": 1303
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 3.214895995706713e-05,
+      "loss": 1.504,
+      "step": 1304
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.2136586740427374e-05,
+      "loss": 1.659,
+      "step": 1305
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.2124206166644404e-05,
+      "loss": 1.7804,
+      "step": 1306
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.211181824322328e-05,
+      "loss": 1.74,
+      "step": 1307
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.20994229776735e-05,
+      "loss": 1.5494,
+      "step": 1308
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.2087020377509026e-05,
+      "loss": 1.5153,
+      "step": 1309
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.2074610450248244e-05,
+      "loss": 1.7559,
+      "step": 1310
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.2062193203414e-05,
+      "loss": 1.5696,
+      "step": 1311
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.204976864453357e-05,
+      "loss": 1.9034,
+      "step": 1312
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.2037336781138674e-05,
+      "loss": 1.3515,
+      "step": 1313
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.202489762076545e-05,
+      "loss": 1.6844,
+      "step": 1314
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.201245117095445e-05,
+      "loss": 1.958,
+      "step": 1315
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.199999743925065e-05,
+      "loss": 1.5395,
+      "step": 1316
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.1987536433203464e-05,
+      "loss": 1.3324,
+      "step": 1317
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.1975068160366685e-05,
+      "loss": 1.5122,
+      "step": 1318
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.1962592628298516e-05,
+      "loss": 2.0091,
+      "step": 1319
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.1950109844561585e-05,
+      "loss": 1.7584,
+      "step": 1320
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.193761981672289e-05,
+      "loss": 1.6667,
+      "step": 1321
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.192512255235382e-05,
+      "loss": 1.4212,
+      "step": 1322
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.191261805903017e-05,
+      "loss": 1.4692,
+      "step": 1323
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.190010634433211e-05,
+      "loss": 1.4009,
+      "step": 1324
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.188758741584417e-05,
+      "loss": 1.8117,
+      "step": 1325
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.1875061281155276e-05,
+      "loss": 1.5054,
+      "step": 1326
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.186252794785872e-05,
+      "loss": 1.2286,
+      "step": 1327
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.1849987423552137e-05,
+      "loss": 2.042,
+      "step": 1328
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.183743971583754e-05,
+      "loss": 1.6218,
+      "step": 1329
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.182488483232129e-05,
+      "loss": 1.8436,
+      "step": 1330
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.1812322780614106e-05,
+      "loss": 1.4028,
+      "step": 1331
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.1799753568331045e-05,
+      "loss": 1.7877,
+      "step": 1332
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.17871772030915e-05,
+      "loss": 1.2856,
+      "step": 1333
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.1774593692519216e-05,
+      "loss": 1.5434,
+      "step": 1334
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.176200304424225e-05,
+      "loss": 1.7546,
+      "step": 1335
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.1749405265893e-05,
+      "loss": 1.7297,
+      "step": 1336
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.1736800365108176e-05,
+      "loss": 1.6379,
+      "step": 1337
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.1724188349528814e-05,
+      "loss": 1.4836,
+      "step": 1338
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.1711569226800266e-05,
+      "loss": 1.1437,
+      "step": 1339
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.1698943004572175e-05,
+      "loss": 2.0096,
+      "step": 1340
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.1686309690498514e-05,
+      "loss": 1.7865,
+      "step": 1341
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.167366929223753e-05,
+      "loss": 1.2965,
+      "step": 1342
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.1661021817451776e-05,
+      "loss": 1.5635,
+      "step": 1343
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.1648367273808086e-05,
+      "loss": 1.3598,
+      "step": 1344
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.1635705668977595e-05,
+      "loss": 1.3299,
+      "step": 1345
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3.162303701063571e-05,
+      "loss": 1.4486,
+      "step": 1346
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.161036130646211e-05,
+      "loss": 1.6958,
+      "step": 1347
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3.159767856414075e-05,
+      "loss": 1.774,
+      "step": 1348
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.1584988791359854e-05,
+      "loss": 1.6047,
+      "step": 1349
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.1572291995811904e-05,
+      "loss": 1.6756,
+      "step": 1350
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3.1559588185193626e-05,
+      "loss": 1.6864,
+      "step": 1351
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.1546877367206026e-05,
+      "loss": 1.6398,
+      "step": 1352
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.153415954955434e-05,
+      "loss": 1.5645,
+      "step": 1353
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3.152143473994805e-05,
+      "loss": 1.5724,
+      "step": 1354
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.150870294610087e-05,
+      "loss": 1.9654,
+      "step": 1355
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.149596417573077e-05,
+      "loss": 1.5469,
+      "step": 1356
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.148321843655993e-05,
+      "loss": 1.4224,
+      "step": 1357
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3.1470465736314744e-05,
+      "loss": 1.4377,
+      "step": 1358
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.145770608272585e-05,
+      "loss": 1.6567,
+      "step": 1359
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.1444939483528095e-05,
+      "loss": 1.6397,
+      "step": 1360
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3.143216594646054e-05,
+      "loss": 1.8574,
+      "step": 1361
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.1419385479266426e-05,
+      "loss": 1.2775,
+      "step": 1362
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3.1406598089693226e-05,
+      "loss": 1.478,
+      "step": 1363
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.139380378549259e-05,
+      "loss": 1.9004,
+      "step": 1364
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.1381002574420376e-05,
+      "loss": 1.4939,
+      "step": 1365
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3.13681944642366e-05,
+      "loss": 1.5028,
+      "step": 1366
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.135537946270551e-05,
+      "loss": 1.4229,
+      "step": 1367
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.1342557577595476e-05,
+      "loss": 1.7939,
+      "step": 1368
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.132972881667908e-05,
+      "loss": 1.8203,
+      "step": 1369
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.1316893187733034e-05,
+      "loss": 1.6721,
+      "step": 1370
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.1304050698538274e-05,
+      "loss": 1.3967,
+      "step": 1371
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3.1291201356879835e-05,
+      "loss": 1.1579,
+      "step": 1372
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.127834517054693e-05,
+      "loss": 1.5373,
+      "step": 1373
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.126548214733292e-05,
+      "loss": 1.6112,
+      "step": 1374
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3.125261229503533e-05,
+      "loss": 1.6061,
+      "step": 1375
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.1239735621455775e-05,
+      "loss": 1.6267,
+      "step": 1376
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3.122685213440006e-05,
+      "loss": 1.7706,
+      "step": 1377
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.1213961841678076e-05,
+      "loss": 1.4984,
+      "step": 1378
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.1201064751103874e-05,
+      "loss": 1.6107,
+      "step": 1379
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.11881608704956e-05,
+      "loss": 1.6984,
+      "step": 1380
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3.1175250207675554e-05,
+      "loss": 1.6016,
+      "step": 1381
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.116233277047008e-05,
+      "loss": 1.5628,
+      "step": 1382
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.1149408566709696e-05,
+      "loss": 1.4395,
+      "step": 1383
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3.113647760422899e-05,
+      "loss": 1.3389,
+      "step": 1384
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.112353989086665e-05,
+      "loss": 1.1154,
+      "step": 1385
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.111059543446545e-05,
+      "loss": 1.9553,
+      "step": 1386
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.1097644242872283e-05,
+      "loss": 1.7059,
+      "step": 1387
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.108468632393808e-05,
+      "loss": 1.5588,
+      "step": 1388
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3.107172168551788e-05,
+      "loss": 1.7193,
+      "step": 1389
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.105875033547079e-05,
+      "loss": 1.644,
+      "step": 1390
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.104577228165998e-05,
+      "loss": 1.3253,
+      "step": 1391
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3.103278753195269e-05,
+      "loss": 1.3949,
+      "step": 1392
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.1019796094220204e-05,
+      "loss": 1.7633,
+      "step": 1393
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.100679797633789e-05,
+      "loss": 1.621,
+      "step": 1394
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3.099379318618514e-05,
+      "loss": 2.134,
+      "step": 1395
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.0980781731645384e-05,
+      "loss": 1.884,
+      "step": 1396
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.096776362060612e-05,
+      "loss": 1.87,
+      "step": 1397
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3.0954738860958875e-05,
+      "loss": 1.2849,
+      "step": 1398
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.0941707460599184e-05,
+      "loss": 1.5124,
+      "step": 1399
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.0928669427426624e-05,
+      "loss": 1.7865,
+      "step": 1400
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3.091562476934479e-05,
+      "loss": 1.5238,
+      "step": 1401
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.090257349426131e-05,
+      "loss": 1.5729,
+      "step": 1402
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.088951561008779e-05,
+      "loss": 1.6424,
+      "step": 1403
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3.0876451124739866e-05,
+      "loss": 1.4362,
+      "step": 1404
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.0863380046137165e-05,
+      "loss": 1.2421,
+      "step": 1405
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.085030238220331e-05,
+      "loss": 1.864,
+      "step": 1406
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3.083721814086594e-05,
+      "loss": 1.4467,
+      "step": 1407
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.082412733005665e-05,
+      "loss": 1.203,
+      "step": 1408
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.081102995771103e-05,
+      "loss": 1.9387,
+      "step": 1409
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3.079792603176864e-05,
+      "loss": 1.5678,
+      "step": 1410
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.078481556017303e-05,
+      "loss": 1.4177,
+      "step": 1411
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.0771698550871696e-05,
+      "loss": 1.5995,
+      "step": 1412
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3.075857501181613e-05,
+      "loss": 1.6175,
+      "step": 1413
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.074544495096175e-05,
+      "loss": 1.4647,
+      "step": 1414
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.0732308376267925e-05,
+      "loss": 1.3895,
+      "step": 1415
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.0719165295698e-05,
+      "loss": 1.5165,
+      "step": 1416
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.070601571721925e-05,
+      "loss": 1.6894,
+      "step": 1417
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.06928596488029e-05,
+      "loss": 1.5155,
+      "step": 1418
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.067969709842407e-05,
+      "loss": 1.2376,
+      "step": 1419
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.066652807406186e-05,
+      "loss": 1.4224,
+      "step": 1420
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.065335258369926e-05,
+      "loss": 1.1845,
+      "step": 1421
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.0640170635323205e-05,
+      "loss": 1.4512,
+      "step": 1422
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.062698223692453e-05,
+      "loss": 1.5051,
+      "step": 1423
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.0613787396497965e-05,
+      "loss": 1.4584,
+      "step": 1424
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.060058612204217e-05,
+      "loss": 1.3538,
+      "step": 1425
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.05873784215597e-05,
+      "loss": 1.5007,
+      "step": 1426
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.057416430305701e-05,
+      "loss": 1.8022,
+      "step": 1427
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3.056094377454443e-05,
+      "loss": 1.7035,
+      "step": 1428
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.054771684403618e-05,
+      "loss": 1.9023,
+      "step": 1429
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3.053448351955036e-05,
+      "loss": 1.6449,
+      "step": 1430
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.052124380910897e-05,
+      "loss": 1.3993,
+      "step": 1431
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.0507997720737853e-05,
+      "loss": 1.4631,
+      "step": 1432
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3.049474526246672e-05,
+      "loss": 1.3959,
+      "step": 1433
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.0481486442329154e-05,
+      "loss": 1.6238,
+      "step": 1434
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.0468221268362587e-05,
+      "loss": 1.786,
+      "step": 1435
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.0454949748608325e-05,
+      "loss": 2.2241,
+      "step": 1436
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.0441671891111482e-05,
+      "loss": 2.0098,
+      "step": 1437
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.0428387703921038e-05,
+      "loss": 1.4205,
+      "step": 1438
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.0415097195089812e-05,
+      "loss": 1.5555,
+      "step": 1439
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.0401800372674447e-05,
+      "loss": 1.468,
+      "step": 1440
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.038849724473542e-05,
+      "loss": 1.5615,
+      "step": 1441
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.0375187819337014e-05,
+      "loss": 1.6855,
+      "step": 1442
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.0361872104547357e-05,
+      "loss": 1.6033,
+      "step": 1443
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.0348550108438367e-05,
+      "loss": 1.9218,
+      "step": 1444
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.033522183908577e-05,
+      "loss": 1.4431,
+      "step": 1445
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.032188730456911e-05,
+      "loss": 1.4761,
+      "step": 1446
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.030854651297172e-05,
+      "loss": 1.7098,
+      "step": 1447
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.029519947238073e-05,
+      "loss": 1.5268,
+      "step": 1448
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.0281846190887045e-05,
+      "loss": 1.5359,
+      "step": 1449
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.0268486676585373e-05,
+      "loss": 1.5894,
+      "step": 1450
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.0255120937574182e-05,
+      "loss": 1.7717,
+      "step": 1451
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.0241748981955728e-05,
+      "loss": 1.7207,
+      "step": 1452
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.022837081783603e-05,
+      "loss": 1.5801,
+      "step": 1453
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.021498645332486e-05,
+      "loss": 1.7173,
+      "step": 1454
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.0201595896535764e-05,
+      "loss": 1.7181,
+      "step": 1455
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.0188199155586048e-05,
+      "loss": 1.9829,
+      "step": 1456
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.0174796238596733e-05,
+      "loss": 1.6401,
+      "step": 1457
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.0161387153692617e-05,
+      "loss": 1.5101,
+      "step": 1458
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.0147971909002227e-05,
+      "loss": 1.5102,
+      "step": 1459
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.0134550512657826e-05,
+      "loss": 1.4212,
+      "step": 1460
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.012112297279539e-05,
+      "loss": 1.3702,
+      "step": 1461
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.0107689297554644e-05,
+      "loss": 1.4332,
+      "step": 1462
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.009424949507902e-05,
+      "loss": 1.5063,
+      "step": 1463
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.008080357351565e-05,
+      "loss": 1.3647,
+      "step": 1464
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.006735154101541e-05,
+      "loss": 1.5842,
+      "step": 1465
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.005389340573285e-05,
+      "loss": 2.0368,
+      "step": 1466
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.0040429175826226e-05,
+      "loss": 1.6111,
+      "step": 1467
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.0026958859457505e-05,
+      "loss": 1.421,
+      "step": 1468
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.0013482464792322e-05,
+      "loss": 2.0083,
+      "step": 1469
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 1.4851,
+      "step": 1470
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 2.9986511473253562e-05,
+      "loss": 1.5824,
+      "step": 1471
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 2.997301689272968e-05,
+      "loss": 1.358,
+      "step": 1472
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.9959516266608713e-05,
+      "loss": 1.4498,
+      "step": 1473
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 2.9946009603074674e-05,
+      "loss": 1.4716,
+      "step": 1474
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 2.9932496910315244e-05,
+      "loss": 1.5452,
+      "step": 1475
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 2.9918978196521756e-05,
+      "loss": 1.3342,
+      "step": 1476
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 2.9905453469889185e-05,
+      "loss": 1.6131,
+      "step": 1477
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.9891922738616163e-05,
+      "loss": 1.8837,
+      "step": 1478
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.987838601090496e-05,
+      "loss": 1.4536,
+      "step": 1479
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.9864843294961474e-05,
+      "loss": 1.5848,
+      "step": 1480
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.9851294598995236e-05,
+      "loss": 1.4877,
+      "step": 1481
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.983773993121941e-05,
+      "loss": 1.689,
+      "step": 1482
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.9824179299850775e-05,
+      "loss": 1.5468,
+      "step": 1483
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.981061271310971e-05,
+      "loss": 1.9664,
+      "step": 1484
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.979704017922023e-05,
+      "loss": 1.4939,
+      "step": 1485
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.9783461706409937e-05,
+      "loss": 1.8347,
+      "step": 1486
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.9769877302910046e-05,
+      "loss": 1.7088,
+      "step": 1487
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.975628697695535e-05,
+      "loss": 1.4696,
+      "step": 1488
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.9742690736784245e-05,
+      "loss": 1.248,
+      "step": 1489
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.9729088590638714e-05,
+      "loss": 1.3557,
+      "step": 1490
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.971548054676432e-05,
+      "loss": 1.4838,
+      "step": 1491
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.9701866613410176e-05,
+      "loss": 1.3048,
+      "step": 1492
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.968824679882901e-05,
+      "loss": 1.5502,
+      "step": 1493
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.9674621111277076e-05,
+      "loss": 1.7122,
+      "step": 1494
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.9660989559014206e-05,
+      "loss": 1.4379,
+      "step": 1495
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 2.9647352150303793e-05,
+      "loss": 1.6113,
+      "step": 1496
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 2.9633708893412764e-05,
+      "loss": 1.4616,
+      "step": 1497
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.9620059796611592e-05,
+      "loss": 1.7133,
+      "step": 1498
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.9606404868174313e-05,
+      "loss": 1.7687,
+      "step": 1499
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 2.959274411637847e-05,
+      "loss": 1.6719,
+      "step": 1500
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 2.9579077549505154e-05,
+      "loss": 1.5018,
+      "step": 1501
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.9565405175838968e-05,
+      "loss": 1.8755,
+      "step": 1502
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.9551727003668053e-05,
+      "loss": 1.4655,
+      "step": 1503
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.9538043041284042e-05,
+      "loss": 1.3982,
+      "step": 1504
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.9524353296982104e-05,
+      "loss": 1.6578,
+      "step": 1505
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.9510657779060885e-05,
+      "loss": 1.5044,
+      "step": 1506
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.9496956495822555e-05,
+      "loss": 1.9118,
+      "step": 1507
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.9483249455572767e-05,
+      "loss": 1.4875,
+      "step": 1508
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.946953666662066e-05,
+      "loss": 1.4675,
+      "step": 1509
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9455818137278876e-05,
+      "loss": 1.5288,
+      "step": 1510
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.9442093875863514e-05,
+      "loss": 1.3117,
+      "step": 1511
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 2.9428363890694158e-05,
+      "loss": 1.3702,
+      "step": 1512
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.9414628190093873e-05,
+      "loss": 1.7517,
+      "step": 1513
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.940088678238916e-05,
+      "loss": 1.5704,
+      "step": 1514
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 2.9387139675910014e-05,
+      "loss": 1.8321,
+      "step": 1515
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.937338687898986e-05,
+      "loss": 1.8364,
+      "step": 1516
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.9359628399965586e-05,
+      "loss": 1.7139,
+      "step": 1517
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 2.934586424717751e-05,
+      "loss": 1.4631,
+      "step": 1518
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.93320944289694e-05,
+      "loss": 1.6186,
+      "step": 1519
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 2.9318318953688462e-05,
+      "loss": 1.3705,
+      "step": 1520
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.9304537829685324e-05,
+      "loss": 1.3648,
+      "step": 1521
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.9290751065314043e-05,
+      "loss": 1.9086,
+      "step": 1522
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 2.9276958668932086e-05,
+      "loss": 1.4185,
+      "step": 1523
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.9263160648900345e-05,
+      "loss": 1.7537,
+      "step": 1524
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.9249357013583112e-05,
+      "loss": 1.6047,
+      "step": 1525
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 2.9235547771348094e-05,
+      "loss": 1.4802,
+      "step": 1526
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.9221732930566384e-05,
+      "loss": 1.588,
+      "step": 1527
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.9207912499612475e-05,
+      "loss": 1.4808,
+      "step": 1528
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 2.919408648686425e-05,
+      "loss": 1.854,
+      "step": 1529
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.9180254900702984e-05,
+      "loss": 1.483,
+      "step": 1530
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.9166417749513308e-05,
+      "loss": 1.3541,
+      "step": 1531
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.915257504168324e-05,
+      "loss": 1.8394,
+      "step": 1532
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 2.9138726785604174e-05,
+      "loss": 1.5394,
+      "step": 1533
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.912487298967086e-05,
+      "loss": 1.4188,
+      "step": 1534
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.9111013662281396e-05,
+      "loss": 1.4069,
+      "step": 1535
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.909714881183725e-05,
+      "loss": 1.6838,
+      "step": 1536
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.9083278446743235e-05,
+      "loss": 1.2903,
+      "step": 1537
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.9069402575407488e-05,
+      "loss": 1.331,
+      "step": 1538
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.9055521206241512e-05,
+      "loss": 1.7254,
+      "step": 1539
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.9041634347660116e-05,
+      "loss": 1.8885,
+      "step": 1540
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.902774200808146e-05,
+      "loss": 1.6041,
+      "step": 1541
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.9013844195927022e-05,
+      "loss": 1.8924,
+      "step": 1542
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.899994091962158e-05,
+      "loss": 1.6669,
+      "step": 1543
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 2.8986032187593248e-05,
+      "loss": 1.7254,
+      "step": 1544
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.897211800827342e-05,
+      "loss": 1.4119,
+      "step": 1545
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 2.895819839009682e-05,
+      "loss": 1.5981,
+      "step": 1546
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.8944273341501454e-05,
+      "loss": 1.8192,
+      "step": 1547
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.893034287092862e-05,
+      "loss": 1.4659,
+      "step": 1548
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 2.8916406986822904e-05,
+      "loss": 1.6989,
+      "step": 1549
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.890246569763218e-05,
+      "loss": 1.8925,
+      "step": 1550
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.8888519011807593e-05,
+      "loss": 1.5884,
+      "step": 1551
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 2.887456693780356e-05,
+      "loss": 1.5611,
+      "step": 1552
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.886060948407776e-05,
+      "loss": 1.5263,
+      "step": 1553
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.8846646659091137e-05,
+      "loss": 1.8506,
+      "step": 1554
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.8832678471307898e-05,
+      "loss": 1.7232,
+      "step": 1555
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 2.8818704929195496e-05,
+      "loss": 1.4397,
+      "step": 1556
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.880472604122462e-05,
+      "loss": 1.1249,
+      "step": 1557
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 2.8790741815869207e-05,
+      "loss": 1.4744,
+      "step": 1558
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.8776752261606435e-05,
+      "loss": 0.9482,
+      "step": 1559
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.876275738691671e-05,
+      "loss": 1.7406,
+      "step": 1560
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.8748757200283655e-05,
+      "loss": 1.6261,
+      "step": 1561
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.8734751710194118e-05,
+      "loss": 1.5121,
+      "step": 1562
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.8720740925138167e-05,
+      "loss": 1.5371,
+      "step": 1563
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.870672485360908e-05,
+      "loss": 1.3179,
+      "step": 1564
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.8692703504103326e-05,
+      "loss": 1.5304,
+      "step": 1565
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.867867688512059e-05,
+      "loss": 1.5685,
+      "step": 1566
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 2.866464500516374e-05,
+      "loss": 1.6405,
+      "step": 1567
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.8650607872738834e-05,
+      "loss": 1.8781,
+      "step": 1568
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.8636565496355128e-05,
+      "loss": 1.4278,
+      "step": 1569
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.8622517884525033e-05,
+      "loss": 1.3788,
+      "step": 1570
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.860846504576416e-05,
+      "loss": 1.6547,
+      "step": 1571
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 2.8594406988591267e-05,
+      "loss": 1.3905,
+      "step": 1572
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.8580343721528278e-05,
+      "loss": 1.6054,
+      "step": 1573
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.85662752531003e-05,
+      "loss": 1.5766,
+      "step": 1574
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 2.8552201591835554e-05,
+      "loss": 1.4084,
+      "step": 1575
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.853812274626544e-05,
+      "loss": 1.1464,
+      "step": 1576
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.852403872492449e-05,
+      "loss": 1.6842,
+      "step": 1577
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.850994953635037e-05,
+      "loss": 1.7176,
+      "step": 1578
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 2.849585518908388e-05,
+      "loss": 1.546,
+      "step": 1579
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.8481755691668947e-05,
+      "loss": 1.8171,
+      "step": 1580
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.846765105265263e-05,
+      "loss": 1.7656,
+      "step": 1581
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 2.8453541280585088e-05,
+      "loss": 1.8665,
+      "step": 1582
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.8439426384019607e-05,
+      "loss": 1.5238,
+      "step": 1583
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.8425306371512567e-05,
+      "loss": 1.4563,
+      "step": 1584
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 2.8411181251623458e-05,
+      "loss": 1.5494,
+      "step": 1585
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.8397051032914862e-05,
+      "loss": 1.5237,
+      "step": 1586
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 2.838291572395245e-05,
+      "loss": 1.6867,
+      "step": 1587
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.8368775333304985e-05,
+      "loss": 1.9004,
+      "step": 1588
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.8354629869544305e-05,
+      "loss": 1.4827,
+      "step": 1589
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 2.8340479341245326e-05,
+      "loss": 1.4123,
+      "step": 1590
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.8326323756986043e-05,
+      "loss": 1.3497,
+      "step": 1591
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.8312163125347485e-05,
+      "loss": 1.2612,
+      "step": 1592
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 2.8297997454913777e-05,
+      "loss": 1.6225,
+      "step": 1593
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.8283826754272082e-05,
+      "loss": 1.5571,
+      "step": 1594
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.8269651032012615e-05,
+      "loss": 1.4949,
+      "step": 1595
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 2.8255470296728623e-05,
+      "loss": 1.4195,
+      "step": 1596
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.824128455701641e-05,
+      "loss": 1.3615,
+      "step": 1597
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.8227093821475317e-05,
+      "loss": 1.7301,
+      "step": 1598
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 2.821289809870768e-05,
+      "loss": 1.4009,
+      "step": 1599
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.81986973973189e-05,
+      "loss": 1.5344,
+      "step": 1600
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.818449172591737e-05,
+      "loss": 1.5542,
+      "step": 1601
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 2.8170281093114503e-05,
+      "loss": 1.6279,
+      "step": 1602
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.8156065507524717e-05,
+      "loss": 1.7285,
+      "step": 1603
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.8141844977765438e-05,
+      "loss": 1.7747,
+      "step": 1604
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 2.812761951245709e-05,
+      "loss": 1.5914,
+      "step": 1605
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.8113389120223077e-05,
+      "loss": 1.6746,
+      "step": 1606
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.8099153809689804e-05,
+      "loss": 1.7096,
+      "step": 1607
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 2.8084913589486645e-05,
+      "loss": 1.3314,
+      "step": 1608
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.8070668468245954e-05,
+      "loss": 1.3847,
+      "step": 1609
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 2.8056418454603068e-05,
+      "loss": 1.7571,
+      "step": 1610
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.8042163557196277e-05,
+      "loss": 1.5367,
+      "step": 1611
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.802790378466683e-05,
+      "loss": 1.6048,
+      "step": 1612
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 2.801363914565894e-05,
+      "loss": 1.6093,
+      "step": 1613
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.7999369648819757e-05,
+      "loss": 1.6822,
+      "step": 1614
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.7985095302799395e-05,
+      "loss": 1.3436,
+      "step": 1615
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 2.7970816116250888e-05,
+      "loss": 1.4577,
+      "step": 1616
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.795653209783023e-05,
+      "loss": 1.6025,
+      "step": 1617
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.794224325619631e-05,
+      "loss": 1.3238,
+      "step": 1618
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 2.7927949600010965e-05,
+      "loss": 1.9677,
+      "step": 1619
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.791365113793894e-05,
+      "loss": 1.4827,
+      "step": 1620
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.78993478786479e-05,
+      "loss": 1.3387,
+      "step": 1621
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.7885039830808422e-05,
+      "loss": 1.7363,
+      "step": 1622
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 2.787072700309397e-05,
+      "loss": 1.7237,
+      "step": 1623
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.785640940418091e-05,
+      "loss": 1.6862,
+      "step": 1624
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 2.784208704274852e-05,
+      "loss": 1.6143,
+      "step": 1625
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.782775992747894e-05,
+      "loss": 1.5145,
+      "step": 1626
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.7813428067057197e-05,
+      "loss": 1.7306,
+      "step": 1627
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 2.779909147017121e-05,
+      "loss": 1.7631,
+      "step": 1628
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.7784750145511755e-05,
+      "loss": 1.7365,
+      "step": 1629
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.7770404101772474e-05,
+      "loss": 1.1499,
+      "step": 1630
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 2.7756053347649872e-05,
+      "loss": 1.6727,
+      "step": 1631
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.7741697891843312e-05,
+      "loss": 1.5976,
+      "step": 1632
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.7727337743055014e-05,
+      "loss": 1.4029,
+      "step": 1633
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 2.7712972909990017e-05,
+      "loss": 1.5816,
+      "step": 1634
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.769860340135623e-05,
+      "loss": 1.6574,
+      "step": 1635
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.768422922586438e-05,
+      "loss": 1.5378,
+      "step": 1636
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 2.7669850392228023e-05,
+      "loss": 1.4194,
+      "step": 1637
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.7655466909163546e-05,
+      "loss": 1.4219,
+      "step": 1638
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 2.7641078785390153e-05,
+      "loss": 1.517,
+      "step": 1639
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.7626686029629852e-05,
+      "loss": 1.7643,
+      "step": 1640
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.7612288650607467e-05,
+      "loss": 1.8053,
+      "step": 1641
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.7597886657050624e-05,
+      "loss": 1.5549,
+      "step": 1642
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 2.7583480057689747e-05,
+      "loss": 1.3846,
+      "step": 1643
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.7569068861258048e-05,
+      "loss": 1.1913,
+      "step": 1644
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.7554653076491534e-05,
+      "loss": 1.2836,
+      "step": 1645
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 2.7540232712128983e-05,
+      "loss": 1.6498,
+      "step": 1646
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.7525807776911954e-05,
+      "loss": 1.5615,
+      "step": 1647
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.751137827958477e-05,
+      "loss": 1.9458,
+      "step": 1648
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 2.7496944228894538e-05,
+      "loss": 1.852,
+      "step": 1649
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.7482505633591104e-05,
+      "loss": 1.926,
+      "step": 1650
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 2.7468062502427084e-05,
+      "loss": 1.8322,
+      "step": 1651
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.7453614844157842e-05,
+      "loss": 1.712,
+      "step": 1652
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.7439162667541464e-05,
+      "loss": 1.8813,
+      "step": 1653
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 2.7424705981338812e-05,
+      "loss": 1.6955,
+      "step": 1654
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.7410244794313455e-05,
+      "loss": 1.2707,
+      "step": 1655
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.73957791152317e-05,
+      "loss": 1.7101,
+      "step": 1656
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 2.7381308952862572e-05,
+      "loss": 1.5946,
+      "step": 1657
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.7366834315977818e-05,
+      "loss": 1.3521,
+      "step": 1658
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.7352355213351905e-05,
+      "loss": 1.6821,
+      "step": 1659
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 2.7337871653761993e-05,
+      "loss": 1.3848,
+      "step": 1660
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.732338364598794e-05,
+      "loss": 1.3858,
+      "step": 1661
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 2.730889119881232e-05,
+      "loss": 1.5035,
+      "step": 1662
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.7294394321020386e-05,
+      "loss": 1.5744,
+      "step": 1663
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.7279893021400084e-05,
+      "loss": 1.9378,
+      "step": 1664
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.7265387308742025e-05,
+      "loss": 1.417,
+      "step": 1665
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 2.7250877191839512e-05,
+      "loss": 1.3923,
+      "step": 1666
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.7236362679488514e-05,
+      "loss": 1.6487,
+      "step": 1667
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.722184378048765e-05,
+      "loss": 1.9384,
+      "step": 1668
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 2.7207320503638216e-05,
+      "loss": 1.5226,
+      "step": 1669
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.719279285774416e-05,
+      "loss": 1.7228,
+      "step": 1670
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.7178260851612076e-05,
+      "loss": 1.3033,
+      "step": 1671
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 2.7163724494051192e-05,
+      "loss": 1.5936,
+      "step": 1672
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.7149183793873388e-05,
+      "loss": 1.6491,
+      "step": 1673
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.713463875989316e-05,
+      "loss": 1.3484,
+      "step": 1674
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 2.7120089400927648e-05,
+      "loss": 1.4672,
+      "step": 1675
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.710553572579661e-05,
+      "loss": 1.8155,
+      "step": 1676
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 2.7090977743322405e-05,
+      "loss": 1.8479,
+      "step": 1677
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.707641546233003e-05,
+      "loss": 1.7194,
+      "step": 1678
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2.7061848891647067e-05,
+      "loss": 1.8594,
+      "step": 1679
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.7047278040103697e-05,
+      "loss": 1.7655,
+      "step": 1680
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 2.7032702916532714e-05,
+      "loss": 1.3717,
+      "step": 1681
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 2.7018123529769484e-05,
+      "loss": 1.4317,
+      "step": 1682
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 2.700353988865197e-05,
+      "loss": 1.1716,
+      "step": 1683
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 2.6988952002020696e-05,
+      "loss": 2.1537,
+      "step": 1684
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 2.6974359878718788e-05,
+      "loss": 1.1254,
+      "step": 1685
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 2.695976352759192e-05,
+      "loss": 1.7918,
+      "step": 1686
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.6945162957488327e-05,
+      "loss": 1.6522,
+      "step": 1687
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 2.6930558177258813e-05,
+      "loss": 1.4893,
+      "step": 1688
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 2.6915949195756726e-05,
+      "loss": 1.4253,
+      "step": 1689
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 2.6901336021837964e-05,
+      "loss": 1.6555,
+      "step": 1690
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 2.6886718664360965e-05,
+      "loss": 1.2829,
+      "step": 1691
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 2.6872097132186706e-05,
+      "loss": 1.9875,
+      "step": 1692
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 2.68574714341787e-05,
+      "loss": 1.5834,
+      "step": 1693
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 2.684284157920296e-05,
+      "loss": 1.6265,
+      "step": 1694
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 2.682820757612805e-05,
+      "loss": 1.7599,
+      "step": 1695
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 2.6813569433825036e-05,
+      "loss": 1.3512,
+      "step": 1696
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 2.6798927161167485e-05,
+      "loss": 1.6642,
+      "step": 1697
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 2.6784280767031487e-05,
+      "loss": 1.628,
+      "step": 1698
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 2.6769630260295607e-05,
+      "loss": 1.8643,
+      "step": 1699
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 2.6754975649840918e-05,
+      "loss": 1.7993,
+      "step": 1700
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 2.6740316944550982e-05,
+      "loss": 1.3498,
+      "step": 1701
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 2.672565415331184e-05,
+      "loss": 1.7265,
+      "step": 1702
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 2.6710987285011997e-05,
+      "loss": 1.4754,
+      "step": 1703
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 2.6696316348542458e-05,
+      "loss": 1.2835,
+      "step": 1704
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 2.668164135279666e-05,
+      "loss": 1.706,
+      "step": 1705
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 2.6666962306670535e-05,
+      "loss": 1.3785,
+      "step": 1706
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 2.6652279219062444e-05,
+      "loss": 1.5208,
+      "step": 1707
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 2.6637592098873203e-05,
+      "loss": 1.458,
+      "step": 1708
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 2.6622900955006086e-05,
+      "loss": 1.5303,
+      "step": 1709
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 2.660820579636679e-05,
+      "loss": 1.5979,
+      "step": 1710
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 2.659350663186346e-05,
+      "loss": 1.5936,
+      "step": 1711
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 2.6578803470406658e-05,
+      "loss": 1.4012,
+      "step": 1712
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 2.656409632090938e-05,
+      "loss": 1.9755,
+      "step": 1713
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 2.6549385192287015e-05,
+      "loss": 1.5753,
+      "step": 1714
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 2.65346700934574e-05,
+      "loss": 1.8329,
+      "step": 1715
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 2.6519951033340748e-05,
+      "loss": 1.5728,
+      "step": 1716
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 2.6505228020859696e-05,
+      "loss": 1.5016,
+      "step": 1717
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 2.6490501064939257e-05,
+      "loss": 1.7003,
+      "step": 1718
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 2.647577017450685e-05,
+      "loss": 1.8286,
+      "step": 1719
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 2.6461035358492274e-05,
+      "loss": 1.3084,
+      "step": 1720
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 2.6446296625827705e-05,
+      "loss": 1.562,
+      "step": 1721
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 2.643155398544769e-05,
+      "loss": 1.6793,
+      "step": 1722
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 2.6416807446289158e-05,
+      "loss": 1.6305,
+      "step": 1723
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 2.6402057017291386e-05,
+      "loss": 1.2797,
+      "step": 1724
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 2.6387302707396025e-05,
+      "loss": 1.6257,
+      "step": 1725
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 2.6372544525547063e-05,
+      "loss": 1.3059,
+      "step": 1726
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 2.6357782480690846e-05,
+      "loss": 1.3164,
+      "step": 1727
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 2.6343016581776057e-05,
+      "loss": 1.4881,
+      "step": 1728
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 2.6328246837753714e-05,
+      "loss": 1.6044,
+      "step": 1729
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 2.631347325757717e-05,
+      "loss": 1.356,
+      "step": 1730
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 2.62986958502021e-05,
+      "loss": 2.0638,
+      "step": 1731
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 2.6283914624586506e-05,
+      "loss": 1.4992,
+      "step": 1732
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 2.6269129589690696e-05,
+      "loss": 1.5455,
+      "step": 1733
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 2.6254340754477284e-05,
+      "loss": 1.81,
+      "step": 1734
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 2.6239548127911204e-05,
+      "loss": 1.9442,
+      "step": 1735
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 2.622475171895967e-05,
+      "loss": 1.9617,
+      "step": 1736
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 2.6209951536592205e-05,
+      "loss": 1.5643,
+      "step": 1737
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 2.61951475897806e-05,
+      "loss": 1.6152,
+      "step": 1738
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 2.618033988749895e-05,
+      "loss": 1.7617,
+      "step": 1739
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 2.616552843872362e-05,
+      "loss": 1.3747,
+      "step": 1740
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 2.6150713252433232e-05,
+      "loss": 1.9039,
+      "step": 1741
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 2.6135894337608686e-05,
+      "loss": 1.3537,
+      "step": 1742
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 2.612107170323314e-05,
+      "loss": 1.512,
+      "step": 1743
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 2.610624535829201e-05,
+      "loss": 1.3189,
+      "step": 1744
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 2.609141531177296e-05,
+      "loss": 1.475,
+      "step": 1745
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 2.6076581572665886e-05,
+      "loss": 1.3685,
+      "step": 1746
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 2.606174414996295e-05,
+      "loss": 1.4686,
+      "step": 1747
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 2.6046903052658515e-05,
+      "loss": 1.4739,
+      "step": 1748
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 2.603205828974919e-05,
+      "loss": 1.8164,
+      "step": 1749
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 2.60172098702338e-05,
+      "loss": 1.4202,
+      "step": 1750
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 2.6002357803113394e-05,
+      "loss": 1.6427,
+      "step": 1751
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 2.5987502097391225e-05,
+      "loss": 1.4448,
+      "step": 1752
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 2.5972642762072752e-05,
+      "loss": 1.8262,
+      "step": 1753
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 2.595777980616564e-05,
+      "loss": 1.3074,
+      "step": 1754
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 2.594291323867974e-05,
+      "loss": 1.5473,
+      "step": 1755
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 2.5928043068627098e-05,
+      "loss": 1.4296,
+      "step": 1756
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 2.5913169305021944e-05,
+      "loss": 1.4966,
+      "step": 1757
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 2.5898291956880687e-05,
+      "loss": 1.4606,
+      "step": 1758
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 2.58834110332219e-05,
+      "loss": 1.4515,
+      "step": 1759
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 2.586852654306634e-05,
+      "loss": 1.5778,
+      "step": 1760
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 2.585363849543691e-05,
+      "loss": 1.4456,
+      "step": 1761
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 2.5838746899358678e-05,
+      "loss": 1.6925,
+      "step": 1762
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 2.582385176385886e-05,
+      "loss": 1.5388,
+      "step": 1763
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 2.5808953097966826e-05,
+      "loss": 1.4287,
+      "step": 1764
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 2.579405091071407e-05,
+      "loss": 1.6265,
+      "step": 1765
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 2.5779145211134233e-05,
+      "loss": 1.4946,
+      "step": 1766
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 2.5764236008263086e-05,
+      "loss": 1.4959,
+      "step": 1767
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 2.574932331113851e-05,
+      "loss": 1.2979,
+      "step": 1768
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 2.573440712880052e-05,
+      "loss": 1.6402,
+      "step": 1769
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 2.5719487470291237e-05,
+      "loss": 1.8233,
+      "step": 1770
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 2.5704564344654887e-05,
+      "loss": 1.8345,
+      "step": 1771
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 2.56896377609378e-05,
+      "loss": 1.4848,
+      "step": 1772
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 2.567470772818841e-05,
+      "loss": 1.6502,
+      "step": 1773
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 2.5659774255457226e-05,
+      "loss": 1.4327,
+      "step": 1774
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 2.5644837351796856e-05,
+      "loss": 1.5404,
+      "step": 1775
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 2.5629897026261974e-05,
+      "loss": 1.6025,
+      "step": 1776
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 2.5614953287909344e-05,
+      "loss": 1.6004,
+      "step": 1777
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 2.5600006145797794e-05,
+      "loss": 1.4519,
+      "step": 1778
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 2.5585055608988202e-05,
+      "loss": 1.6914,
+      "step": 1779
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 2.5570101686543528e-05,
+      "loss": 1.3179,
+      "step": 1780
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 2.555514438752877e-05,
+      "loss": 1.5372,
+      "step": 1781
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 2.5540183721010962e-05,
+      "loss": 1.3886,
+      "step": 1782
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 2.5525219696059197e-05,
+      "loss": 1.5602,
+      "step": 1783
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 2.5510252321744605e-05,
+      "loss": 1.3773,
+      "step": 1784
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 2.549528160714033e-05,
+      "loss": 1.7017,
+      "step": 1785
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 2.548030756132156e-05,
+      "loss": 1.9173,
+      "step": 1786
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 2.5465330193365483e-05,
+      "loss": 1.7219,
+      "step": 1787
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 2.5450349512351324e-05,
+      "loss": 1.5566,
+      "step": 1788
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 2.543536552736029e-05,
+      "loss": 1.3496,
+      "step": 1789
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 2.5420378247475608e-05,
+      "loss": 1.769,
+      "step": 1790
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 2.5405387681782498e-05,
+      "loss": 1.8872,
+      "step": 1791
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 2.5390393839368174e-05,
+      "loss": 1.9834,
+      "step": 1792
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 2.5375396729321832e-05,
+      "loss": 1.4948,
+      "step": 1793
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 2.5360396360734664e-05,
+      "loss": 1.682,
+      "step": 1794
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 2.5345392742699804e-05,
+      "loss": 1.4626,
+      "step": 1795
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 2.5330385884312385e-05,
+      "loss": 1.629,
+      "step": 1796
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 2.53153757946695e-05,
+      "loss": 1.4507,
+      "step": 1797
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 2.5300362482870192e-05,
+      "loss": 1.7575,
+      "step": 1798
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 2.5285345958015466e-05,
+      "loss": 1.5675,
+      "step": 1799
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 2.5270326229208263e-05,
+      "loss": 1.1872,
+      "step": 1800
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 2.5255303305553485e-05,
+      "loss": 1.5548,
+      "step": 1801
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 2.5240277196157947e-05,
+      "loss": 1.4136,
+      "step": 1802
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 2.5225247910130412e-05,
+      "loss": 1.6415,
+      "step": 1803
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 2.521021545658157e-05,
+      "loss": 1.487,
+      "step": 1804
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 2.5195179844624018e-05,
+      "loss": 1.4505,
+      "step": 1805
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 2.5180141083372283e-05,
+      "loss": 1.4225,
+      "step": 1806
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 2.5165099181942783e-05,
+      "loss": 1.5531,
+      "step": 1807
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 2.5150054149453853e-05,
+      "loss": 1.5176,
+      "step": 1808
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 2.5135005995025727e-05,
+      "loss": 1.659,
+      "step": 1809
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 2.5119954727780528e-05,
+      "loss": 1.756,
+      "step": 1810
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 2.5104900356842256e-05,
+      "loss": 1.7891,
+      "step": 1811
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 2.5089842891336805e-05,
+      "loss": 1.806,
+      "step": 1812
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.5074782340391947e-05,
+      "loss": 1.3335,
+      "step": 1813
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.505971871313732e-05,
+      "loss": 1.502,
+      "step": 1814
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.5044652018704415e-05,
+      "loss": 1.4413,
+      "step": 1815
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.5029582266226605e-05,
+      "loss": 1.5881,
+      "step": 1816
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.5014509464839095e-05,
+      "loss": 1.6972,
+      "step": 1817
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.499943362367895e-05,
+      "loss": 1.6799,
+      "step": 1818
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.498435475188508e-05,
+      "loss": 1.2557,
+      "step": 1819
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.4969272858598225e-05,
+      "loss": 1.6154,
+      "step": 1820
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.4954187952960967e-05,
+      "loss": 1.4868,
+      "step": 1821
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.4939100044117694e-05,
+      "loss": 1.5371,
+      "step": 1822
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.492400914121463e-05,
+      "loss": 1.6533,
+      "step": 1823
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.490891525339982e-05,
+      "loss": 1.5346,
+      "step": 1824
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.4893818389823104e-05,
+      "loss": 1.8092,
+      "step": 1825
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.4878718559636127e-05,
+      "loss": 1.3466,
+      "step": 1826
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.4863615771992345e-05,
+      "loss": 1.6715,
+      "step": 1827
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.4848510036046993e-05,
+      "loss": 1.3495,
+      "step": 1828
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.48334013609571e-05,
+      "loss": 1.2564,
+      "step": 1829
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.4818289755881474e-05,
+      "loss": 1.6498,
+      "step": 1830
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.4803175229980706e-05,
+      "loss": 1.2582,
+      "step": 1831
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.4788057792417147e-05,
+      "loss": 1.9926,
+      "step": 1832
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.4772937452354916e-05,
+      "loss": 1.7435,
+      "step": 1833
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.4757814218959904e-05,
+      "loss": 1.7305,
+      "step": 1834
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.4742688101399744e-05,
+      "loss": 1.3161,
+      "step": 1835
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.472755910884381e-05,
+      "loss": 1.6769,
+      "step": 1836
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 2.471242725046323e-05,
+      "loss": 1.3568,
+      "step": 1837
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 2.4697292535430867e-05,
+      "loss": 1.5298,
+      "step": 1838
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 2.4682154972921325e-05,
+      "loss": 1.7617,
+      "step": 1839
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 2.466701457211092e-05,
+      "loss": 1.3976,
+      "step": 1840
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 2.4651871342177686e-05,
+      "loss": 1.6639,
+      "step": 1841
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 2.463672529230139e-05,
+      "loss": 1.8387,
+      "step": 1842
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 2.4621576431663486e-05,
+      "loss": 1.3803,
+      "step": 1843
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 2.4606424769447163e-05,
+      "loss": 1.5526,
+      "step": 1844
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 2.459127031483727e-05,
+      "loss": 1.6956,
+      "step": 1845
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 2.4576113077020373e-05,
+      "loss": 1.4062,
+      "step": 1846
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 2.4560953065184724e-05,
+      "loss": 1.5567,
+      "step": 1847
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 2.4545790288520256e-05,
+      "loss": 1.5245,
+      "step": 1848
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 2.4530624756218568e-05,
+      "loss": 1.8041,
+      "step": 1849
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 2.4515456477472935e-05,
+      "loss": 1.3402,
+      "step": 1850
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 2.45002854614783e-05,
+      "loss": 1.6466,
+      "step": 1851
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 2.4485111717431266e-05,
+      "loss": 2.1338,
+      "step": 1852
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 2.4469935254530084e-05,
+      "loss": 1.3905,
+      "step": 1853
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 2.4454756081974654e-05,
+      "loss": 1.8913,
+      "step": 1854
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 2.443957420896653e-05,
+      "loss": 1.3724,
+      "step": 1855
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 2.4424389644708888e-05,
+      "loss": 1.4121,
+      "step": 1856
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 2.4409202398406536e-05,
+      "loss": 1.5653,
+      "step": 1857
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 2.439401247926592e-05,
+      "loss": 1.6762,
+      "step": 1858
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 2.4378819896495097e-05,
+      "loss": 1.7609,
+      "step": 1859
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 2.4363624659303738e-05,
+      "loss": 1.6285,
+      "step": 1860
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 2.4348426776903132e-05,
+      "loss": 1.5825,
+      "step": 1861
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 2.433322625850617e-05,
+      "loss": 1.9076,
+      "step": 1862
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 2.431802311332732e-05,
+      "loss": 1.4476,
+      "step": 1863
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 2.430281735058267e-05,
+      "loss": 1.2678,
+      "step": 1864
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 2.4287608979489882e-05,
+      "loss": 1.3049,
+      "step": 1865
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 2.4272398009268203e-05,
+      "loss": 1.5188,
+      "step": 1866
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 2.425718444913845e-05,
+      "loss": 1.6143,
+      "step": 1867
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 2.424196830832301e-05,
+      "loss": 1.26,
+      "step": 1868
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 2.422674959604585e-05,
+      "loss": 1.6625,
+      "step": 1869
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 2.421152832153247e-05,
+      "loss": 1.5021,
+      "step": 1870
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 2.4196304494009945e-05,
+      "loss": 1.49,
+      "step": 1871
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 2.4181078122706886e-05,
+      "loss": 1.8062,
+      "step": 1872
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 2.4165849216853454e-05,
+      "loss": 1.4773,
+      "step": 1873
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 2.4150617785681333e-05,
+      "loss": 1.3187,
+      "step": 1874
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 2.4135383838423758e-05,
+      "loss": 1.6306,
+      "step": 1875
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 2.4120147384315465e-05,
+      "loss": 1.8806,
+      "step": 1876
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 2.4104908432592732e-05,
+      "loss": 1.5697,
+      "step": 1877
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.408966699249334e-05,
+      "loss": 1.5744,
+      "step": 1878
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 2.407442307325658e-05,
+      "loss": 1.3193,
+      "step": 1879
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 2.4059176684123246e-05,
+      "loss": 1.4727,
+      "step": 1880
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 2.4043927834335624e-05,
+      "loss": 1.7232,
+      "step": 1881
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 2.4028676533137506e-05,
+      "loss": 1.6639,
+      "step": 1882
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 2.4013422789774147e-05,
+      "loss": 1.7433,
+      "step": 1883
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 2.399816661349231e-05,
+      "loss": 1.3445,
+      "step": 1884
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 2.3982908013540214e-05,
+      "loss": 1.6695,
+      "step": 1885
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 2.3967646999167552e-05,
+      "loss": 1.6012,
+      "step": 1886
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.395238357962548e-05,
+      "loss": 1.7213,
+      "step": 1887
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.3937117764166614e-05,
+      "loss": 1.8832,
+      "step": 1888
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.392184956204503e-05,
+      "loss": 1.2114,
+      "step": 1889
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.3906578982516225e-05,
+      "loss": 1.4155,
+      "step": 1890
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.3891306034837163e-05,
+      "loss": 1.6668,
+      "step": 1891
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.3876030728266235e-05,
+      "loss": 1.6266,
+      "step": 1892
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.3860753072063264e-05,
+      "loss": 1.4303,
+      "step": 1893
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.3845473075489487e-05,
+      "loss": 1.5783,
+      "step": 1894
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.3830190747807578e-05,
+      "loss": 1.8303,
+      "step": 1895
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.3814906098281613e-05,
+      "loss": 2.1199,
+      "step": 1896
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.3799619136177067e-05,
+      "loss": 1.3604,
+      "step": 1897
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.3784329870760835e-05,
+      "loss": 1.3638,
+      "step": 1898
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.3769038311301193e-05,
+      "loss": 1.8488,
+      "step": 1899
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.3753744467067817e-05,
+      "loss": 1.2473,
+      "step": 1900
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.373844834733177e-05,
+      "loss": 1.5829,
+      "step": 1901
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.3723149961365482e-05,
+      "loss": 1.5595,
+      "step": 1902
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.3707849318442776e-05,
+      "loss": 1.276,
+      "step": 1903
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.369254642783881e-05,
+      "loss": 1.9508,
+      "step": 1904
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.367724129883015e-05,
+      "loss": 1.67,
+      "step": 1905
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.3661933940694677e-05,
+      "loss": 1.5106,
+      "step": 1906
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.3646624362711655e-05,
+      "loss": 1.5551,
+      "step": 1907
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.3631312574161674e-05,
+      "loss": 1.478,
+      "step": 1908
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.3615998584326668e-05,
+      "loss": 1.4786,
+      "step": 1909
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.3600682402489907e-05,
+      "loss": 1.4091,
+      "step": 1910
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.3585364037935995e-05,
+      "loss": 1.5968,
+      "step": 1911
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.357004349995085e-05,
+      "loss": 1.4592,
+      "step": 1912
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.355472079782172e-05,
+      "loss": 1.3018,
+      "step": 1913
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.3539395940837143e-05,
+      "loss": 1.6292,
+      "step": 1914
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.3524068938286984e-05,
+      "loss": 1.6617,
+      "step": 1915
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.350873979946241e-05,
+      "loss": 1.522,
+      "step": 1916
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.3493408533655852e-05,
+      "loss": 1.8138,
+      "step": 1917
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.3478075150161076e-05,
+      "loss": 1.3671,
+      "step": 1918
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.34627396582731e-05,
+      "loss": 1.5916,
+      "step": 1919
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.3447402067288227e-05,
+      "loss": 1.4515,
+      "step": 1920
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.3432062386504036e-05,
+      "loss": 1.7122,
+      "step": 1921
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.3416720625219373e-05,
+      "loss": 1.6524,
+      "step": 1922
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.3401376792734344e-05,
+      "loss": 1.4418,
+      "step": 1923
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.33860308983503e-05,
+      "loss": 1.6078,
+      "step": 1924
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.337068295136987e-05,
+      "loss": 1.4268,
+      "step": 1925
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.3355332961096888e-05,
+      "loss": 1.5037,
+      "step": 1926
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.3339980936836467e-05,
+      "loss": 1.3935,
+      "step": 1927
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.332462688789492e-05,
+      "loss": 2.0469,
+      "step": 1928
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.3309270823579803e-05,
+      "loss": 1.6362,
+      "step": 1929
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.3293912753199905e-05,
+      "loss": 1.6072,
+      "step": 1930
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.3278552686065194e-05,
+      "loss": 1.6738,
+      "step": 1931
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.3263190631486895e-05,
+      "loss": 1.5493,
+      "step": 1932
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.32478265987774e-05,
+      "loss": 1.2538,
+      "step": 1933
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.323246059725032e-05,
+      "loss": 1.75,
+      "step": 1934
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.321709263622046e-05,
+      "loss": 1.6096,
+      "step": 1935
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.3201722725003796e-05,
+      "loss": 1.8876,
+      "step": 1936
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.3186350872917514e-05,
+      "loss": 1.4209,
+      "step": 1937
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.3170977089279945e-05,
+      "loss": 1.568,
+      "step": 1938
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.3155601383410614e-05,
+      "loss": 1.9411,
+      "step": 1939
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.31402237646302e-05,
+      "loss": 1.6221,
+      "step": 1940
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.3124844242260543e-05,
+      "loss": 1.5811,
+      "step": 1941
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.3109462825624646e-05,
+      "loss": 1.2129,
+      "step": 1942
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.3094079524046652e-05,
+      "loss": 1.7817,
+      "step": 1943
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.3078694346851843e-05,
+      "loss": 1.7809,
+      "step": 1944
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.306330730336664e-05,
+      "loss": 1.6608,
+      "step": 1945
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.3047918402918607e-05,
+      "loss": 1.5364,
+      "step": 1946
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.3032527654836425e-05,
+      "loss": 1.4671,
+      "step": 1947
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.3017135068449884e-05,
+      "loss": 1.5417,
+      "step": 1948
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.3001740653089907e-05,
+      "loss": 1.4274,
+      "step": 1949
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.2986344418088518e-05,
+      "loss": 1.4833,
+      "step": 1950
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.2970946372778847e-05,
+      "loss": 1.4972,
+      "step": 1951
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.295554652649511e-05,
+      "loss": 1.4087,
+      "step": 1952
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.2940144888572627e-05,
+      "loss": 1.8596,
+      "step": 1953
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.2924741468347803e-05,
+      "loss": 1.8281,
+      "step": 1954
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.2909336275158117e-05,
+      "loss": 1.6063,
+      "step": 1955
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.289392931834213e-05,
+      "loss": 1.2814,
+      "step": 1956
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.2878520607239464e-05,
+      "loss": 1.3046,
+      "step": 1957
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.2863110151190816e-05,
+      "loss": 1.6494,
+      "step": 1958
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.2847697959537927e-05,
+      "loss": 1.0714,
+      "step": 1959
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.2832284041623597e-05,
+      "loss": 1.5849,
+      "step": 1960
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.2816868406791673e-05,
+      "loss": 1.662,
+      "step": 1961
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.2801451064387048e-05,
+      "loss": 1.5935,
+      "step": 1962
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.2786032023755635e-05,
+      "loss": 1.414,
+      "step": 1963
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.27706112942444e-05,
+      "loss": 1.4471,
+      "step": 1964
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.27551888852013e-05,
+      "loss": 1.5038,
+      "step": 1965
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.2739764805975343e-05,
+      "loss": 1.5838,
+      "step": 1966
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.272433906591652e-05,
+      "loss": 1.6108,
+      "step": 1967
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.2708911674375864e-05,
+      "loss": 1.5713,
+      "step": 1968
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.2693482640705377e-05,
+      "loss": 2.1799,
+      "step": 1969
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.2678051974258067e-05,
+      "loss": 1.8421,
+      "step": 1970
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.2662619684387938e-05,
+      "loss": 1.613,
+      "step": 1971
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.2647185780449972e-05,
+      "loss": 1.549,
+      "step": 1972
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.2631750271800126e-05,
+      "loss": 1.6281,
+      "step": 1973
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.2616313167795344e-05,
+      "loss": 1.4721,
+      "step": 1974
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.2600874477793516e-05,
+      "loss": 1.5875,
+      "step": 1975
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.2585434211153515e-05,
+      "loss": 1.7286,
+      "step": 1976
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.2569992377235163e-05,
+      "loss": 1.2174,
+      "step": 1977
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.255454898539921e-05,
+      "loss": 1.4628,
+      "step": 1978
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.253910404500739e-05,
+      "loss": 1.3871,
+      "step": 1979
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.2523657565422344e-05,
+      "loss": 1.5449,
+      "step": 1980
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.250820955600766e-05,
+      "loss": 1.7845,
+      "step": 1981
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.249276002612785e-05,
+      "loss": 1.9458,
+      "step": 1982
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.2477308985148348e-05,
+      "loss": 1.7966,
+      "step": 1983
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.2461856442435502e-05,
+      "loss": 1.3651,
+      "step": 1984
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.2446402407356577e-05,
+      "loss": 1.4341,
+      "step": 1985
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.2430946889279737e-05,
+      "loss": 1.4448,
+      "step": 1986
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.2415489897574047e-05,
+      "loss": 1.2733,
+      "step": 1987
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.2400031441609454e-05,
+      "loss": 1.0902,
+      "step": 1988
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.2384571530756814e-05,
+      "loss": 1.6331,
+      "step": 1989
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.2369110174387854e-05,
+      "loss": 1.56,
+      "step": 1990
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.235364738187518e-05,
+      "loss": 1.919,
+      "step": 1991
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.2338183162592256e-05,
+      "loss": 1.4982,
+      "step": 1992
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.232271752591342e-05,
+      "loss": 1.4311,
+      "step": 1993
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.2307250481213884e-05,
+      "loss": 1.3946,
+      "step": 1994
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.2291782037869684e-05,
+      "loss": 1.7442,
+      "step": 1995
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.227631220525773e-05,
+      "loss": 1.2936,
+      "step": 1996
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.2260840992755765e-05,
+      "loss": 1.8694,
+      "step": 1997
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.2245368409742357e-05,
+      "loss": 1.7031,
+      "step": 1998
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.222989446559692e-05,
+      "loss": 1.4699,
+      "step": 1999
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.221441916969969e-05,
+      "loss": 1.5158,
+      "step": 2000
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.2198942531431714e-05,
+      "loss": 1.1724,
+      "step": 2001
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.2183464560174865e-05,
+      "loss": 1.7464,
+      "step": 2002
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.2167985265311818e-05,
+      "loss": 1.5075,
+      "step": 2003
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.2152504656226053e-05,
+      "loss": 1.6495,
+      "step": 2004
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.2137022742301838e-05,
+      "loss": 1.3268,
+      "step": 2005
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.2121539532924238e-05,
+      "loss": 1.5035,
+      "step": 2006
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.2106055037479102e-05,
+      "loss": 1.6231,
+      "step": 2007
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.209056926535307e-05,
+      "loss": 1.7987,
+      "step": 2008
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.207508222593354e-05,
+      "loss": 1.4956,
+      "step": 2009
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.205959392860868e-05,
+      "loss": 1.655,
+      "step": 2010
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.204410438276744e-05,
+      "loss": 1.428,
+      "step": 2011
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.202861359779949e-05,
+      "loss": 1.4285,
+      "step": 2012
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.201312158309529e-05,
+      "loss": 1.7496,
+      "step": 2013
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.1997628348046015e-05,
+      "loss": 1.6075,
+      "step": 2014
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.1982133902043605e-05,
+      "loss": 1.3263,
+      "step": 2015
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.1966638254480713e-05,
+      "loss": 1.6539,
+      "step": 2016
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.195114141475074e-05,
+      "loss": 1.9489,
+      "step": 2017
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.193564339224779e-05,
+      "loss": 1.6455,
+      "step": 2018
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.192014419636669e-05,
+      "loss": 1.5185,
+      "step": 2019
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.1904643836502997e-05,
+      "loss": 1.5523,
+      "step": 2020
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.188914232205294e-05,
+      "loss": 1.2684,
+      "step": 2021
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.1873639662413473e-05,
+      "loss": 1.5666,
+      "step": 2022
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.1858135866982243e-05,
+      "loss": 1.8229,
+      "step": 2023
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.1842630945157572e-05,
+      "loss": 1.5063,
+      "step": 2024
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.1827124906338483e-05,
+      "loss": 1.6609,
+      "step": 2025
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.181161775992465e-05,
+      "loss": 1.5321,
+      "step": 2026
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.1796109515316445e-05,
+      "loss": 1.8181,
+      "step": 2027
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.1780600181914887e-05,
+      "loss": 1.6743,
+      "step": 2028
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.1765089769121674e-05,
+      "loss": 1.5287,
+      "step": 2029
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.1749578286339138e-05,
+      "loss": 1.3871,
+      "step": 2030
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.1734065742970274e-05,
+      "loss": 2.1786,
+      "step": 2031
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.1718552148418707e-05,
+      "loss": 1.4472,
+      "step": 2032
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.1703037512088713e-05,
+      "loss": 1.4796,
+      "step": 2033
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.1687521843385192e-05,
+      "loss": 1.5038,
+      "step": 2034
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.167200515171367e-05,
+      "loss": 1.7611,
+      "step": 2035
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.1656487446480295e-05,
+      "loss": 1.2831,
+      "step": 2036
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.1640968737091832e-05,
+      "loss": 1.3912,
+      "step": 2037
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.1625449032955652e-05,
+      "loss": 1.8937,
+      "step": 2038
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.1609928343479725e-05,
+      "loss": 1.6109,
+      "step": 2039
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.1594406678072616e-05,
+      "loss": 1.3756,
+      "step": 2040
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.1578884046143496e-05,
+      "loss": 1.4513,
+      "step": 2041
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.156336045710211e-05,
+      "loss": 1.5382,
+      "step": 2042
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.1547835920358787e-05,
+      "loss": 1.4344,
+      "step": 2043
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.1532310445324432e-05,
+      "loss": 1.628,
+      "step": 2044
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.1516784041410516e-05,
+      "loss": 1.6818,
+      "step": 2045
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.150125671802907e-05,
+      "loss": 1.6018,
+      "step": 2046
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.148572848459269e-05,
+      "loss": 1.9328,
+      "step": 2047
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.1470199350514513e-05,
+      "loss": 1.8448,
+      "step": 2048
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.1454669325208236e-05,
+      "loss": 1.5391,
+      "step": 2049
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.1439138418088094e-05,
+      "loss": 1.331,
+      "step": 2050
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.1423606638568844e-05,
+      "loss": 1.7967,
+      "step": 2051
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.140807399606578e-05,
+      "loss": 1.4829,
+      "step": 2052
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.1392540499994718e-05,
+      "loss": 1.71,
+      "step": 2053
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.1377006159771988e-05,
+      "loss": 1.4711,
+      "step": 2054
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.136147098481445e-05,
+      "loss": 1.4984,
+      "step": 2055
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.1345934984539444e-05,
+      "loss": 1.4128,
+      "step": 2056
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.1330398168364823e-05,
+      "loss": 1.6092,
+      "step": 2057
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.1314860545708944e-05,
+      "loss": 1.9022,
+      "step": 2058
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.1299322125990623e-05,
+      "loss": 1.6861,
+      "step": 2059
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.1283782918629195e-05,
+      "loss": 1.8001,
+      "step": 2060
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.126824293304445e-05,
+      "loss": 1.7529,
+      "step": 2061
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.1252702178656656e-05,
+      "loss": 2.1322,
+      "step": 2062
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.123716066488655e-05,
+      "loss": 1.7293,
+      "step": 2063
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.1221618401155325e-05,
+      "loss": 1.4073,
+      "step": 2064
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.120607539688463e-05,
+      "loss": 1.3429,
+      "step": 2065
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.1190531661496565e-05,
+      "loss": 1.9693,
+      "step": 2066
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.1174987204413662e-05,
+      "loss": 1.5199,
+      "step": 2067
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.115944203505891e-05,
+      "loss": 1.572,
+      "step": 2068
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.1143896162855713e-05,
+      "loss": 1.7873,
+      "step": 2069
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.1128349597227915e-05,
+      "loss": 1.4575,
+      "step": 2070
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.1112802347599764e-05,
+      "loss": 1.9235,
+      "step": 2071
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.1097254423395937e-05,
+      "loss": 1.49,
+      "step": 2072
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.1081705834041507e-05,
+      "loss": 1.3529,
+      "step": 2073
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.106615658896197e-05,
+      "loss": 1.4954,
+      "step": 2074
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.1050606697583197e-05,
+      "loss": 1.5733,
+      "step": 2075
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.103505616933146e-05,
+      "loss": 1.7504,
+      "step": 2076
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.1019505013633427e-05,
+      "loss": 1.3657,
+      "step": 2077
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.1003953239916125e-05,
+      "loss": 1.438,
+      "step": 2078
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.0988400857606982e-05,
+      "loss": 1.2855,
+      "step": 2079
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.0972847876133776e-05,
+      "loss": 1.7313,
+      "step": 2080
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.095729430492464e-05,
+      "loss": 1.4443,
+      "step": 2081
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.0941740153408095e-05,
+      "loss": 1.4381,
+      "step": 2082
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.0926185431012985e-05,
+      "loss": 1.708,
+      "step": 2083
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.091063014716852e-05,
+      "loss": 1.5783,
+      "step": 2084
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.089507431130424e-05,
+      "loss": 1.5714,
+      "step": 2085
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.0879517932850018e-05,
+      "loss": 1.3688,
+      "step": 2086
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.0863961021236053e-05,
+      "loss": 1.7495,
+      "step": 2087
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.0848403585892886e-05,
+      "loss": 1.3422,
+      "step": 2088
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.083284563625136e-05,
+      "loss": 1.661,
+      "step": 2089
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.0817287181742624e-05,
+      "loss": 1.6731,
+      "step": 2090
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.080172823179815e-05,
+      "loss": 1.3708,
+      "step": 2091
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.0786168795849697e-05,
+      "loss": 1.3586,
+      "step": 2092
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.0770608883329335e-05,
+      "loss": 1.5536,
+      "step": 2093
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.0755048503669397e-05,
+      "loss": 1.4956,
+      "step": 2094
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.073948766630252e-05,
+      "loss": 1.6249,
+      "step": 2095
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.0723926380661612e-05,
+      "loss": 1.4322,
+      "step": 2096
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.070836465617985e-05,
+      "loss": 1.5464,
+      "step": 2097
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.069280250229069e-05,
+      "loss": 1.7179,
+      "step": 2098
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.0677239928427832e-05,
+      "loss": 1.172,
+      "step": 2099
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.0661676944025233e-05,
+      "loss": 1.3673,
+      "step": 2100
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.06461135585171e-05,
+      "loss": 1.6115,
+      "step": 2101
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.0630549781337904e-05,
+      "loss": 1.9836,
+      "step": 2102
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.0614985621922314e-05,
+      "loss": 1.4617,
+      "step": 2103
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.0599421089705264e-05,
+      "loss": 1.6105,
+      "step": 2104
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.05838561941219e-05,
+      "loss": 1.7223,
+      "step": 2105
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.0568290944607596e-05,
+      "loss": 1.4894,
+      "step": 2106
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.0552725350597917e-05,
+      "loss": 1.3827,
+      "step": 2107
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.053715942152867e-05,
+      "loss": 1.5141,
+      "step": 2108
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.052159316683585e-05,
+      "loss": 1.7696,
+      "step": 2109
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.050602659595564e-05,
+      "loss": 1.2083,
+      "step": 2110
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.0490459718324432e-05,
+      "loss": 1.1804,
+      "step": 2111
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.047489254337879e-05,
+      "loss": 1.2626,
+      "step": 2112
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.0459325080555473e-05,
+      "loss": 1.597,
+      "step": 2113
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.0443757339291395e-05,
+      "loss": 1.6255,
+      "step": 2114
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.0428189329023652e-05,
+      "loss": 1.6546,
+      "step": 2115
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.0412621059189503e-05,
+      "loss": 1.6551,
+      "step": 2116
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.0397052539226362e-05,
+      "loss": 1.2773,
+      "step": 2117
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.0381483778571792e-05,
+      "loss": 1.3574,
+      "step": 2118
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.0365914786663512e-05,
+      "loss": 1.6483,
+      "step": 2119
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.035034557293936e-05,
+      "loss": 1.3532,
+      "step": 2120
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.033477614683733e-05,
+      "loss": 1.5954,
+      "step": 2121
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.0319206517795537e-05,
+      "loss": 1.3096,
+      "step": 2122
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.0303636695252217e-05,
+      "loss": 1.4244,
+      "step": 2123
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.0288066688645728e-05,
+      "loss": 1.6646,
+      "step": 2124
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.0272496507414533e-05,
+      "loss": 1.5699,
+      "step": 2125
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.0256926160997203e-05,
+      "loss": 1.3446,
+      "step": 2126
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.024135565883241e-05,
+      "loss": 1.6686,
+      "step": 2127
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.0225785010358933e-05,
+      "loss": 1.7363,
+      "step": 2128
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.0210214225015607e-05,
+      "loss": 1.7121,
+      "step": 2129
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.0194643312241392e-05,
+      "loss": 1.3719,
+      "step": 2130
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.0179072281475287e-05,
+      "loss": 1.5433,
+      "step": 2131
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.016350114215639e-05,
+      "loss": 1.5172,
+      "step": 2132
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.014792990372385e-05,
+      "loss": 1.7098,
+      "step": 2133
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.0132358575616885e-05,
+      "loss": 1.7457,
+      "step": 2134
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.011678716727475e-05,
+      "loss": 1.8878,
+      "step": 2135
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.0101215688136774e-05,
+      "loss": 1.9205,
+      "step": 2136
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.0085644147642315e-05,
+      "loss": 1.6121,
+      "step": 2137
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.0070072555230764e-05,
+      "loss": 1.3761,
+      "step": 2138
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.0054500920341547e-05,
+      "loss": 1.3624,
+      "step": 2139
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.0038929252414136e-05,
+      "loss": 1.7097,
+      "step": 2140
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.002335756088798e-05,
+      "loss": 1.9451,
+      "step": 2141
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.0007785855202577e-05,
+      "loss": 1.5347,
+      "step": 2142
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 1.999221414479743e-05,
+      "loss": 1.5063,
+      "step": 2143
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 1.9976642439112027e-05,
+      "loss": 1.5959,
+      "step": 2144
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 1.996107074758587e-05,
+      "loss": 1.665,
+      "step": 2145
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 1.994549907965845e-05,
+      "loss": 1.5139,
+      "step": 2146
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 1.992992744476924e-05,
+      "loss": 1.47,
+      "step": 2147
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 1.9914355852357688e-05,
+      "loss": 1.8611,
+      "step": 2148
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 1.9898784311863226e-05,
+      "loss": 1.6086,
+      "step": 2149
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 1.988321283272526e-05,
+      "loss": 1.7381,
+      "step": 2150
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 1.986764142438313e-05,
+      "loss": 1.6229,
+      "step": 2151
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 1.985207009627616e-05,
+      "loss": 1.7679,
+      "step": 2152
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 1.983649885784362e-05,
+      "loss": 1.4012,
+      "step": 2153
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 1.982092771852472e-05,
+      "loss": 1.6003,
+      "step": 2154
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 1.9805356687758618e-05,
+      "loss": 1.2749,
+      "step": 2155
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 1.9789785774984396e-05,
+      "loss": 1.3781,
+      "step": 2156
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 1.9774214989641077e-05,
+      "loss": 1.5511,
+      "step": 2157
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 1.9758644341167596e-05,
+      "loss": 1.5551,
+      "step": 2158
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 1.9743073839002807e-05,
+      "loss": 1.6239,
+      "step": 2159
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 1.9727503492585477e-05,
+      "loss": 1.8501,
+      "step": 2160
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 1.971193331135428e-05,
+      "loss": 1.6105,
+      "step": 2161
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 1.9696363304747786e-05,
+      "loss": 1.6567,
+      "step": 2162
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 1.968079348220447e-05,
+      "loss": 1.4222,
+      "step": 2163
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 1.9665223853162673e-05,
+      "loss": 1.6832,
+      "step": 2164
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 1.9649654427060646e-05,
+      "loss": 1.6838,
+      "step": 2165
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 1.9634085213336498e-05,
+      "loss": 1.6298,
+      "step": 2166
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 1.961851622142821e-05,
+      "loss": 1.6966,
+      "step": 2167
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 1.960294746077364e-05,
+      "loss": 1.5156,
+      "step": 2168
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 1.95873789408105e-05,
+      "loss": 1.3088,
+      "step": 2169
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 1.9571810670976355e-05,
+      "loss": 1.5737,
+      "step": 2170
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 1.9556242660708612e-05,
+      "loss": 1.8201,
+      "step": 2171
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 1.9540674919444533e-05,
+      "loss": 1.2383,
+      "step": 2172
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 1.9525107456621214e-05,
+      "loss": 1.4108,
+      "step": 2173
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 1.950954028167557e-05,
+      "loss": 1.5006,
+      "step": 2174
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 1.949397340404436e-05,
+      "loss": 1.6324,
+      "step": 2175
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 1.9478406833164154e-05,
+      "loss": 1.7764,
+      "step": 2176
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 1.9462840578471338e-05,
+      "loss": 1.5894,
+      "step": 2177
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 1.9447274649402093e-05,
+      "loss": 1.3269,
+      "step": 2178
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 1.9431709055392418e-05,
+      "loss": 1.5497,
+      "step": 2179
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 1.9416143805878106e-05,
+      "loss": 1.2604,
+      "step": 2180
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 1.9400578910294743e-05,
+      "loss": 1.3627,
+      "step": 2181
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 1.9385014378077693e-05,
+      "loss": 1.2967,
+      "step": 2182
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 1.9369450218662106e-05,
+      "loss": 1.5703,
+      "step": 2183
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 1.9353886441482903e-05,
+      "loss": 1.3604,
+      "step": 2184
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 1.9338323055974773e-05,
+      "loss": 1.9312,
+      "step": 2185
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 1.9322760071572175e-05,
+      "loss": 1.3547,
+      "step": 2186
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 1.930719749770931e-05,
+      "loss": 1.3831,
+      "step": 2187
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 1.929163534382015e-05,
+      "loss": 1.6534,
+      "step": 2188
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 1.9276073619338395e-05,
+      "loss": 1.4285,
+      "step": 2189
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 1.9260512333697492e-05,
+      "loss": 1.8141,
+      "step": 2190
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 1.9244951496330616e-05,
+      "loss": 1.2735,
+      "step": 2191
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 1.922939111667068e-05,
+      "loss": 1.6,
+      "step": 2192
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 1.921383120415031e-05,
+      "loss": 1.5299,
+      "step": 2193
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 1.9198271768201857e-05,
+      "loss": 1.344,
+      "step": 2194
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 1.9182712818257383e-05,
+      "loss": 1.499,
+      "step": 2195
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 1.916715436374865e-05,
+      "loss": 1.6506,
+      "step": 2196
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 1.915159641410712e-05,
+      "loss": 2.1004,
+      "step": 2197
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 1.913603897876395e-05,
+      "loss": 1.5991,
+      "step": 2198
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 1.912048206714999e-05,
+      "loss": 1.6462,
+      "step": 2199
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 1.9104925688695766e-05,
+      "loss": 1.6169,
+      "step": 2200
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.908936985283148e-05,
+      "loss": 1.4872,
+      "step": 2201
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 1.907381456898701e-05,
+      "loss": 1.5135,
+      "step": 2202
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 1.9058259846591905e-05,
+      "loss": 1.1838,
+      "step": 2203
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 1.904270569507537e-05,
+      "loss": 1.8702,
+      "step": 2204
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 1.9027152123866238e-05,
+      "loss": 1.6824,
+      "step": 2205
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 1.9011599142393025e-05,
+      "loss": 1.5891,
+      "step": 2206
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 1.899604676008388e-05,
+      "loss": 1.4911,
+      "step": 2207
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 1.8980494986366583e-05,
+      "loss": 1.6254,
+      "step": 2208
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 1.8964943830668545e-05,
+      "loss": 1.4295,
+      "step": 2209
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 1.8949393302416813e-05,
+      "loss": 1.3704,
+      "step": 2210
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 1.8933843411038038e-05,
+      "loss": 1.5936,
+      "step": 2211
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 1.8918294165958496e-05,
+      "loss": 1.362,
+      "step": 2212
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 1.890274557660407e-05,
+      "loss": 1.4803,
+      "step": 2213
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 1.888719765240024e-05,
+      "loss": 1.5196,
+      "step": 2214
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 1.887165040277209e-05,
+      "loss": 1.5422,
+      "step": 2215
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 1.8856103837144287e-05,
+      "loss": 1.5012,
+      "step": 2216
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 1.88405579649411e-05,
+      "loss": 1.5574,
+      "step": 2217
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 1.8825012795586348e-05,
+      "loss": 1.4939,
+      "step": 2218
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 1.8809468338503445e-05,
+      "loss": 1.559,
+      "step": 2219
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 1.8793924603115375e-05,
+      "loss": 1.3207,
+      "step": 2220
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 1.877838159884468e-05,
+      "loss": 1.4796,
+      "step": 2221
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 1.8762839335113454e-05,
+      "loss": 1.4068,
+      "step": 2222
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 1.874729782134335e-05,
+      "loss": 1.2925,
+      "step": 2223
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 1.8731757066955557e-05,
+      "loss": 1.2989,
+      "step": 2224
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 1.871621708137081e-05,
+      "loss": 1.3097,
+      "step": 2225
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 1.870067787400938e-05,
+      "loss": 1.2782,
+      "step": 2226
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 1.8685139454291063e-05,
+      "loss": 1.2836,
+      "step": 2227
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 1.8669601831635184e-05,
+      "loss": 1.9537,
+      "step": 2228
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 1.865406501546056e-05,
+      "loss": 1.9396,
+      "step": 2229
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 1.8638529015185553e-05,
+      "loss": 1.7575,
+      "step": 2230
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 1.8622993840228015e-05,
+      "loss": 1.7564,
+      "step": 2231
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 1.8607459500005292e-05,
+      "loss": 1.7836,
+      "step": 2232
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 1.859192600393423e-05,
+      "loss": 1.6943,
+      "step": 2233
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 1.8576393361431162e-05,
+      "loss": 1.3956,
+      "step": 2234
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 1.8560861581911913e-05,
+      "loss": 1.2419,
+      "step": 2235
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 1.8545330674791768e-05,
+      "loss": 1.7248,
+      "step": 2236
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 1.852980064948549e-05,
+      "loss": 1.1733,
+      "step": 2237
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 1.8514271515407318e-05,
+      "loss": 1.797,
+      "step": 2238
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 1.8498743281970936e-05,
+      "loss": 1.5744,
+      "step": 2239
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 1.848321595858949e-05,
+      "loss": 1.7529,
+      "step": 2240
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 1.846768955467557e-05,
+      "loss": 1.7067,
+      "step": 2241
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 1.8452164079641217e-05,
+      "loss": 1.3932,
+      "step": 2242
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 1.8436639542897892e-05,
+      "loss": 1.8114,
+      "step": 2243
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 1.8421115953856504e-05,
+      "loss": 1.6392,
+      "step": 2244
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 1.8405593321927394e-05,
+      "loss": 1.597,
+      "step": 2245
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 1.8390071656520288e-05,
+      "loss": 1.6454,
+      "step": 2246
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 1.8374550967044355e-05,
+      "loss": 1.7114,
+      "step": 2247
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 1.835903126290817e-05,
+      "loss": 1.5321,
+      "step": 2248
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 1.8343512553519708e-05,
+      "loss": 1.4977,
+      "step": 2249
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 1.8327994848286334e-05,
+      "loss": 1.8077,
+      "step": 2250
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 1.8312478156614814e-05,
+      "loss": 1.6962,
+      "step": 2251
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 1.8296962487911293e-05,
+      "loss": 1.656,
+      "step": 2252
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 1.82814478515813e-05,
+      "loss": 1.4892,
+      "step": 2253
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 1.8265934257029733e-05,
+      "loss": 1.6038,
+      "step": 2254
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 1.8250421713660865e-05,
+      "loss": 1.7756,
+      "step": 2255
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 1.823491023087833e-05,
+      "loss": 1.3729,
+      "step": 2256
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 1.8219399818085113e-05,
+      "loss": 1.4655,
+      "step": 2257
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 1.8203890484683565e-05,
+      "loss": 1.4575,
+      "step": 2258
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 1.818838224007536e-05,
+      "loss": 1.5206,
+      "step": 2259
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 1.8172875093661527e-05,
+      "loss": 1.8173,
+      "step": 2260
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 1.815736905484243e-05,
+      "loss": 1.8612,
+      "step": 2261
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 1.814186413301776e-05,
+      "loss": 1.2859,
+      "step": 2262
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 1.812636033758653e-05,
+      "loss": 1.6211,
+      "step": 2263
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 1.8110857677947064e-05,
+      "loss": 1.5788,
+      "step": 2264
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 1.8095356163497013e-05,
+      "loss": 1.617,
+      "step": 2265
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 1.8079855803633317e-05,
+      "loss": 1.5893,
+      "step": 2266
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 1.8064356607752217e-05,
+      "loss": 1.3263,
+      "step": 2267
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 1.8048858585249264e-05,
+      "loss": 1.7777,
+      "step": 2268
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 1.803336174551929e-05,
+      "loss": 1.6854,
+      "step": 2269
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 1.80178660979564e-05,
+      "loss": 1.7025,
+      "step": 2270
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 1.8002371651953985e-05,
+      "loss": 1.9915,
+      "step": 2271
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 1.798687841690472e-05,
+      "loss": 1.8689,
+      "step": 2272
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 1.797138640220052e-05,
+      "loss": 1.5685,
+      "step": 2273
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 1.795589561723257e-05,
+      "loss": 1.6415,
+      "step": 2274
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 1.7940406071391323e-05,
+      "loss": 1.5248,
+      "step": 2275
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 1.7924917774066465e-05,
+      "loss": 1.5908,
+      "step": 2276
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 1.7909430734646936e-05,
+      "loss": 1.9701,
+      "step": 2277
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 1.78939449625209e-05,
+      "loss": 1.5747,
+      "step": 2278
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 1.787846046707577e-05,
+      "loss": 1.3736,
+      "step": 2279
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 1.786297725769817e-05,
+      "loss": 1.7773,
+      "step": 2280
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 1.7847495343773954e-05,
+      "loss": 1.4384,
+      "step": 2281
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 1.7832014734688182e-05,
+      "loss": 1.5591,
+      "step": 2282
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 1.7816535439825135e-05,
+      "loss": 1.4102,
+      "step": 2283
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 1.7801057468568286e-05,
+      "loss": 1.6166,
+      "step": 2284
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 1.7785580830300323e-05,
+      "loss": 1.7626,
+      "step": 2285
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 1.777010553440309e-05,
+      "loss": 1.9002,
+      "step": 2286
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 1.7754631590257653e-05,
+      "loss": 1.2919,
+      "step": 2287
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 1.7739159007244245e-05,
+      "loss": 1.3997,
+      "step": 2288
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 1.7723687794742272e-05,
+      "loss": 1.7025,
+      "step": 2289
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 1.770821796213032e-05,
+      "loss": 1.4776,
+      "step": 2290
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 1.7692749518786123e-05,
+      "loss": 1.5678,
+      "step": 2291
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 1.7677282474086582e-05,
+      "loss": 1.6687,
+      "step": 2292
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 1.766181683740775e-05,
+      "loss": 1.3164,
+      "step": 2293
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 1.7646352618124827e-05,
+      "loss": 1.2713,
+      "step": 2294
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 1.7630889825612146e-05,
+      "loss": 1.742,
+      "step": 2295
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 1.7615428469243185e-05,
+      "loss": 1.4517,
+      "step": 2296
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 1.7599968558390553e-05,
+      "loss": 1.4411,
+      "step": 2297
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 1.7584510102425963e-05,
+      "loss": 1.6968,
+      "step": 2298
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 1.756905311072027e-05,
+      "loss": 1.7024,
+      "step": 2299
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 1.755359759264343e-05,
+      "loss": 1.4962,
+      "step": 2300
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 1.7538143557564504e-05,
+      "loss": 1.7303,
+      "step": 2301
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 1.752269101485166e-05,
+      "loss": 1.2358,
+      "step": 2302
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 1.7507239973872154e-05,
+      "loss": 1.3597,
+      "step": 2303
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 1.7491790443992347e-05,
+      "loss": 1.4719,
+      "step": 2304
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 1.7476342434577662e-05,
+      "loss": 1.6307,
+      "step": 2305
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.7460895954992616e-05,
+      "loss": 1.4644,
+      "step": 2306
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.744545101460079e-05,
+      "loss": 1.6151,
+      "step": 2307
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.7430007622764844e-05,
+      "loss": 1.9879,
+      "step": 2308
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.7414565788846485e-05,
+      "loss": 1.4649,
+      "step": 2309
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.7399125522206484e-05,
+      "loss": 1.4175,
+      "step": 2310
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.738368683220466e-05,
+      "loss": 1.7082,
+      "step": 2311
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.7368249728199884e-05,
+      "loss": 1.8698,
+      "step": 2312
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.735281421955004e-05,
+      "loss": 1.6636,
+      "step": 2313
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.7337380315612072e-05,
+      "loss": 1.892,
+      "step": 2314
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.732194802574194e-05,
+      "loss": 1.6168,
+      "step": 2315
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.7306517359294633e-05,
+      "loss": 1.9203,
+      "step": 2316
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.7291088325624142e-05,
+      "loss": 1.5575,
+      "step": 2317
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.7275660934083484e-05,
+      "loss": 1.5673,
+      "step": 2318
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.7260235194024667e-05,
+      "loss": 1.4435,
+      "step": 2319
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.7244811114798705e-05,
+      "loss": 1.5784,
+      "step": 2320
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.722938870575561e-05,
+      "loss": 1.3913,
+      "step": 2321
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.7213967976244365e-05,
+      "loss": 1.5361,
+      "step": 2322
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.7198548935612955e-05,
+      "loss": 1.411,
+      "step": 2323
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.718313159320833e-05,
+      "loss": 1.5553,
+      "step": 2324
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.7167715958376407e-05,
+      "loss": 1.4439,
+      "step": 2325
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.7152302040462087e-05,
+      "loss": 1.4563,
+      "step": 2326
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.7136889848809194e-05,
+      "loss": 1.5282,
+      "step": 2327
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.7121479392760543e-05,
+      "loss": 1.3193,
+      "step": 2328
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.710607068165788e-05,
+      "loss": 1.7875,
+      "step": 2329
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.709066372484189e-05,
+      "loss": 1.3642,
+      "step": 2330
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.7075258531652204e-05,
+      "loss": 1.343,
+      "step": 2331
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.7059855111427377e-05,
+      "loss": 1.4082,
+      "step": 2332
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.7044453473504896e-05,
+      "loss": 1.4109,
+      "step": 2333
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.702905362722116e-05,
+      "loss": 1.6045,
+      "step": 2334
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.701365558191149e-05,
+      "loss": 1.5478,
+      "step": 2335
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.6998259346910096e-05,
+      "loss": 1.5529,
+      "step": 2336
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.698286493155012e-05,
+      "loss": 1.5769,
+      "step": 2337
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.696747234516358e-05,
+      "loss": 1.5414,
+      "step": 2338
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.6952081597081396e-05,
+      "loss": 1.7613,
+      "step": 2339
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.6936692696633365e-05,
+      "loss": 1.3544,
+      "step": 2340
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.6921305653148167e-05,
+      "loss": 1.4945,
+      "step": 2341
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.6905920475953358e-05,
+      "loss": 1.3867,
+      "step": 2342
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.6890537174375357e-05,
+      "loss": 1.5584,
+      "step": 2343
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.687515575773946e-05,
+      "loss": 1.6609,
+      "step": 2344
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.6859776235369807e-05,
+      "loss": 1.4804,
+      "step": 2345
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.6844398616589393e-05,
+      "loss": 1.5401,
+      "step": 2346
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.6829022910720058e-05,
+      "loss": 1.5925,
+      "step": 2347
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.681364912708249e-05,
+      "loss": 1.7124,
+      "step": 2348
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.67982772749962e-05,
+      "loss": 1.6046,
+      "step": 2349
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.6782907363779543e-05,
+      "loss": 1.402,
+      "step": 2350
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.676753940274968e-05,
+      "loss": 1.7085,
+      "step": 2351
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.6752173401222602e-05,
+      "loss": 2.065,
+      "step": 2352
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.673680936851312e-05,
+      "loss": 1.701,
+      "step": 2353
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.6721447313934816e-05,
+      "loss": 1.085,
+      "step": 2354
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.670608724680011e-05,
+      "loss": 1.5819,
+      "step": 2355
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.6690729176420204e-05,
+      "loss": 1.4256,
+      "step": 2356
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.6675373112105087e-05,
+      "loss": 1.5631,
+      "step": 2357
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.666001906316354e-05,
+      "loss": 1.5447,
+      "step": 2358
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.6644667038903115e-05,
+      "loss": 1.5571,
+      "step": 2359
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.6629317048630137e-05,
+      "loss": 1.6271,
+      "step": 2360
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.6613969101649702e-05,
+      "loss": 1.4735,
+      "step": 2361
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.659862320726566e-05,
+      "loss": 1.4511,
+      "step": 2362
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.6583279374780627e-05,
+      "loss": 1.4263,
+      "step": 2363
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.6567937613495964e-05,
+      "loss": 1.7056,
+      "step": 2364
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.6552597932711773e-05,
+      "loss": 1.6368,
+      "step": 2365
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.653726034172691e-05,
+      "loss": 1.572,
+      "step": 2366
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.6521924849838934e-05,
+      "loss": 1.3875,
+      "step": 2367
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.6506591466344154e-05,
+      "loss": 1.3423,
+      "step": 2368
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.6491260200537605e-05,
+      "loss": 1.8068,
+      "step": 2369
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.6475931061713022e-05,
+      "loss": 1.3217,
+      "step": 2370
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.6460604059162863e-05,
+      "loss": 1.3777,
+      "step": 2371
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.6445279202178287e-05,
+      "loss": 1.339,
+      "step": 2372
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.6429956500049156e-05,
+      "loss": 1.5434,
+      "step": 2373
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.641463596206401e-05,
+      "loss": 1.5853,
+      "step": 2374
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.6399317597510096e-05,
+      "loss": 1.5012,
+      "step": 2375
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.6384001415673335e-05,
+      "loss": 1.5106,
+      "step": 2376
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.636868742583833e-05,
+      "loss": 1.3978,
+      "step": 2377
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.6353375637288345e-05,
+      "loss": 1.4743,
+      "step": 2378
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.633806605930532e-05,
+      "loss": 1.4638,
+      "step": 2379
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.632275870116986e-05,
+      "loss": 1.5421,
+      "step": 2380
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.6307453572161195e-05,
+      "loss": 1.5221,
+      "step": 2381
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.6292150681557237e-05,
+      "loss": 1.4364,
+      "step": 2382
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.6276850038634524e-05,
+      "loss": 1.9469,
+      "step": 2383
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.6261551652668237e-05,
+      "loss": 1.2945,
+      "step": 2384
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.6246255532932186e-05,
+      "loss": 1.4592,
+      "step": 2385
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.623096168869881e-05,
+      "loss": 1.3428,
+      "step": 2386
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.621567012923917e-05,
+      "loss": 1.4557,
+      "step": 2387
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.6200380863822936e-05,
+      "loss": 1.6227,
+      "step": 2388
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.6185093901718393e-05,
+      "loss": 1.4087,
+      "step": 2389
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.6169809252192422e-05,
+      "loss": 1.8408,
+      "step": 2390
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.6154526924510513e-05,
+      "loss": 1.2805,
+      "step": 2391
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.613924692793674e-05,
+      "loss": 1.2557,
+      "step": 2392
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.6123969271733775e-05,
+      "loss": 1.425,
+      "step": 2393
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.6108693965162847e-05,
+      "loss": 1.7334,
+      "step": 2394
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.6093421017483785e-05,
+      "loss": 1.857,
+      "step": 2395
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.607815043795498e-05,
+      "loss": 1.2576,
+      "step": 2396
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.606288223583339e-05,
+      "loss": 1.756,
+      "step": 2397
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.6047616420374526e-05,
+      "loss": 1.4891,
+      "step": 2398
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.6032353000832454e-05,
+      "loss": 1.5835,
+      "step": 2399
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.6017091986459793e-05,
+      "loss": 1.7853,
+      "step": 2400
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.6001833386507693e-05,
+      "loss": 1.5094,
+      "step": 2401
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.5986577210225857e-05,
+      "loss": 1.7919,
+      "step": 2402
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.5971323466862504e-05,
+      "loss": 1.528,
+      "step": 2403
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.5956072165664386e-05,
+      "loss": 1.7151,
+      "step": 2404
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.5940823315876765e-05,
+      "loss": 1.6558,
+      "step": 2405
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.5925576926743426e-05,
+      "loss": 1.6823,
+      "step": 2406
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.5910333007506666e-05,
+      "loss": 1.7588,
+      "step": 2407
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.589509156740727e-05,
+      "loss": 1.3366,
+      "step": 2408
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.5879852615684538e-05,
+      "loss": 1.2968,
+      "step": 2409
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.586461616157625e-05,
+      "loss": 1.5179,
+      "step": 2410
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.5849382214318674e-05,
+      "loss": 1.5159,
+      "step": 2411
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.5834150783146553e-05,
+      "loss": 1.54,
+      "step": 2412
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.5818921877293117e-05,
+      "loss": 1.537,
+      "step": 2413
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.5803695505990058e-05,
+      "loss": 1.68,
+      "step": 2414
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.5788471678467533e-05,
+      "loss": 1.2813,
+      "step": 2415
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.5773250403954153e-05,
+      "loss": 1.6215,
+      "step": 2416
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.575803169167699e-05,
+      "loss": 1.3614,
+      "step": 2417
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.5742815550861554e-05,
+      "loss": 1.3287,
+      "step": 2418
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.57276019907318e-05,
+      "loss": 1.3035,
+      "step": 2419
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.5712391020510118e-05,
+      "loss": 1.9239,
+      "step": 2420
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.569718264941734e-05,
+      "loss": 1.7385,
+      "step": 2421
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.568197688667269e-05,
+      "loss": 1.4137,
+      "step": 2422
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.566677374149384e-05,
+      "loss": 2.0586,
+      "step": 2423
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.565157322309687e-05,
+      "loss": 1.4988,
+      "step": 2424
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.5636375340696265e-05,
+      "loss": 1.6179,
+      "step": 2425
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.5621180103504913e-05,
+      "loss": 1.4203,
+      "step": 2426
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.5605987520734088e-05,
+      "loss": 1.9065,
+      "step": 2427
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.559079760159347e-05,
+      "loss": 1.4011,
+      "step": 2428
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.557561035529112e-05,
+      "loss": 1.703,
+      "step": 2429
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.5560425791033474e-05,
+      "loss": 1.5025,
+      "step": 2430
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.554524391802535e-05,
+      "loss": 1.4549,
+      "step": 2431
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.553006474546992e-05,
+      "loss": 1.5301,
+      "step": 2432
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.5514888282568738e-05,
+      "loss": 1.6419,
+      "step": 2433
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.549971453852171e-05,
+      "loss": 1.3825,
+      "step": 2434
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.5484543522527075e-05,
+      "loss": 1.4531,
+      "step": 2435
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.5469375243781445e-05,
+      "loss": 1.8821,
+      "step": 2436
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.545420971147975e-05,
+      "loss": 1.5753,
+      "step": 2437
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.543904693481528e-05,
+      "loss": 1.6571,
+      "step": 2438
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.542388692297963e-05,
+      "loss": 1.497,
+      "step": 2439
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.5408729685162738e-05,
+      "loss": 1.8412,
+      "step": 2440
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.5393575230552844e-05,
+      "loss": 1.4801,
+      "step": 2441
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.5378423568336517e-05,
+      "loss": 1.4243,
+      "step": 2442
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.5363274707698617e-05,
+      "loss": 1.2185,
+      "step": 2443
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.5348128657822317e-05,
+      "loss": 2.0143,
+      "step": 2444
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.5332985427889087e-05,
+      "loss": 1.5149,
+      "step": 2445
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.531784502707868e-05,
+      "loss": 1.6442,
+      "step": 2446
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.5302707464569132e-05,
+      "loss": 1.5262,
+      "step": 2447
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.5287572749536778e-05,
+      "loss": 1.5204,
+      "step": 2448
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.5272440891156204e-05,
+      "loss": 1.3221,
+      "step": 2449
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.5257311898600266e-05,
+      "loss": 1.4647,
+      "step": 2450
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.5242185781040099e-05,
+      "loss": 1.497,
+      "step": 2451
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.5227062547645086e-05,
+      "loss": 1.6273,
+      "step": 2452
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.521194220758286e-05,
+      "loss": 1.6241,
+      "step": 2453
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.5196824770019301e-05,
+      "loss": 1.5637,
+      "step": 2454
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.518171024411853e-05,
+      "loss": 1.678,
+      "step": 2455
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.5166598639042907e-05,
+      "loss": 1.5265,
+      "step": 2456
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.5151489963953012e-05,
+      "loss": 1.7018,
+      "step": 2457
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.5136384228007658e-05,
+      "loss": 1.4667,
+      "step": 2458
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.5121281440363876e-05,
+      "loss": 1.4195,
+      "step": 2459
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.51061816101769e-05,
+      "loss": 1.7559,
+      "step": 2460
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.5091084746600188e-05,
+      "loss": 1.9077,
+      "step": 2461
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.5075990858785377e-05,
+      "loss": 1.5998,
+      "step": 2462
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.5060899955882316e-05,
+      "loss": 2.1943,
+      "step": 2463
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.5045812047039043e-05,
+      "loss": 1.2843,
+      "step": 2464
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.503072714140178e-05,
+      "loss": 1.3384,
+      "step": 2465
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.5015645248114924e-05,
+      "loss": 1.5228,
+      "step": 2466
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.5000566376321055e-05,
+      "loss": 1.527,
+      "step": 2467
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.4985490535160913e-05,
+      "loss": 1.3497,
+      "step": 2468
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.4970417733773403e-05,
+      "loss": 1.335,
+      "step": 2469
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.4955347981295587e-05,
+      "loss": 1.4879,
+      "step": 2470
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.4940281286862686e-05,
+      "loss": 1.7671,
+      "step": 2471
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.4925217659608053e-05,
+      "loss": 1.7835,
+      "step": 2472
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.49101571086632e-05,
+      "loss": 1.3305,
+      "step": 2473
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.4895099643157751e-05,
+      "loss": 1.5452,
+      "step": 2474
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.4880045272219484e-05,
+      "loss": 1.6827,
+      "step": 2475
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.486499400497428e-05,
+      "loss": 1.6957,
+      "step": 2476
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.4849945850546153e-05,
+      "loss": 1.2977,
+      "step": 2477
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.4834900818057225e-05,
+      "loss": 1.6303,
+      "step": 2478
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.4819858916627726e-05,
+      "loss": 1.4309,
+      "step": 2479
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.4804820155375987e-05,
+      "loss": 1.7732,
+      "step": 2480
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.4789784543418437e-05,
+      "loss": 1.9123,
+      "step": 2481
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.4774752089869593e-05,
+      "loss": 1.6161,
+      "step": 2482
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.4759722803842058e-05,
+      "loss": 1.5111,
+      "step": 2483
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.4744696694446523e-05,
+      "loss": 1.8654,
+      "step": 2484
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.4729673770791739e-05,
+      "loss": 1.2314,
+      "step": 2485
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.4714654041984537e-05,
+      "loss": 1.0682,
+      "step": 2486
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.469963751712981e-05,
+      "loss": 1.3562,
+      "step": 2487
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.468462420533051e-05,
+      "loss": 0.9844,
+      "step": 2488
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.4669614115687625e-05,
+      "loss": 1.4101,
+      "step": 2489
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.4654607257300208e-05,
+      "loss": 1.7272,
+      "step": 2490
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.463960363926535e-05,
+      "loss": 1.3358,
+      "step": 2491
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.4624603270678171e-05,
+      "loss": 1.6289,
+      "step": 2492
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.4609606160631833e-05,
+      "loss": 2.0345,
+      "step": 2493
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.4594612318217509e-05,
+      "loss": 1.5279,
+      "step": 2494
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.45796217525244e-05,
+      "loss": 1.4728,
+      "step": 2495
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.4564634472639717e-05,
+      "loss": 1.8082,
+      "step": 2496
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.4549650487648683e-05,
+      "loss": 1.3325,
+      "step": 2497
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.4534669806634516e-05,
+      "loss": 1.3532,
+      "step": 2498
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.4519692438678444e-05,
+      "loss": 1.4105,
+      "step": 2499
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.4504718392859669e-05,
+      "loss": 1.1649,
+      "step": 2500
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.4489747678255396e-05,
+      "loss": 1.548,
+      "step": 2501
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.447478030394081e-05,
+      "loss": 1.7542,
+      "step": 2502
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.4459816278989046e-05,
+      "loss": 1.8301,
+      "step": 2503
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.4444855612471243e-05,
+      "loss": 1.6121,
+      "step": 2504
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.4429898313456479e-05,
+      "loss": 1.7821,
+      "step": 2505
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.44149443910118e-05,
+      "loss": 1.572,
+      "step": 2506
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.4399993854202214e-05,
+      "loss": 1.5306,
+      "step": 2507
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.4385046712090661e-05,
+      "loss": 1.6272,
+      "step": 2508
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.4370102973738031e-05,
+      "loss": 1.4669,
+      "step": 2509
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.4355162648203151e-05,
+      "loss": 1.6213,
+      "step": 2510
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.4340225744542781e-05,
+      "loss": 1.5625,
+      "step": 2511
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.4325292271811595e-05,
+      "loss": 1.7258,
+      "step": 2512
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.43103622390622e-05,
+      "loss": 1.9316,
+      "step": 2513
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.4295435655345115e-05,
+      "loss": 1.7913,
+      "step": 2514
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4280512529708772e-05,
+      "loss": 1.2026,
+      "step": 2515
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4265592871199486e-05,
+      "loss": 1.8009,
+      "step": 2516
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4250676688861496e-05,
+      "loss": 1.4557,
+      "step": 2517
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4235763991736921e-05,
+      "loss": 1.9124,
+      "step": 2518
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4220854788865772e-05,
+      "loss": 1.7006,
+      "step": 2519
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4205949089285937e-05,
+      "loss": 1.9532,
+      "step": 2520
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4191046902033181e-05,
+      "loss": 1.49,
+      "step": 2521
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4176148236141143e-05,
+      "loss": 1.4806,
+      "step": 2522
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4161253100641328e-05,
+      "loss": 1.4466,
+      "step": 2523
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4146361504563096e-05,
+      "loss": 1.653,
+      "step": 2524
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4131473456933665e-05,
+      "loss": 1.4029,
+      "step": 2525
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4116588966778104e-05,
+      "loss": 1.7652,
+      "step": 2526
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4101708043119318e-05,
+      "loss": 1.9277,
+      "step": 2527
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4086830694978057e-05,
+      "loss": 1.684,
+      "step": 2528
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4071956931372912e-05,
+      "loss": 1.6603,
+      "step": 2529
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4057086761320272e-05,
+      "loss": 1.6757,
+      "step": 2530
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4042220193834368e-05,
+      "loss": 1.521,
+      "step": 2531
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4027357237927253e-05,
+      "loss": 1.3469,
+      "step": 2532
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4012497902608782e-05,
+      "loss": 1.7434,
+      "step": 2533
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.399764219688661e-05,
+      "loss": 1.5187,
+      "step": 2534
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.3982790129766203e-05,
+      "loss": 1.2073,
+      "step": 2535
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.3967941710250817e-05,
+      "loss": 1.4496,
+      "step": 2536
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.3953096947341492e-05,
+      "loss": 1.6661,
+      "step": 2537
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.3938255850037055e-05,
+      "loss": 1.3808,
+      "step": 2538
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.3923418427334113e-05,
+      "loss": 1.3113,
+      "step": 2539
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.3908584688227045e-05,
+      "loss": 1.5003,
+      "step": 2540
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.389375464170799e-05,
+      "loss": 1.5257,
+      "step": 2541
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.3878928296766868e-05,
+      "loss": 1.982,
+      "step": 2542
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.3864105662391327e-05,
+      "loss": 2.0552,
+      "step": 2543
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.3849286747566781e-05,
+      "loss": 1.6354,
+      "step": 2544
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.3834471561276391e-05,
+      "loss": 1.7387,
+      "step": 2545
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.3819660112501054e-05,
+      "loss": 1.2605,
+      "step": 2546
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.3804852410219404e-05,
+      "loss": 2.1124,
+      "step": 2547
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.3790048463407803e-05,
+      "loss": 1.5086,
+      "step": 2548
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.3775248281040334e-05,
+      "loss": 1.5658,
+      "step": 2549
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.3760451872088804e-05,
+      "loss": 1.3903,
+      "step": 2550
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.3745659245522722e-05,
+      "loss": 1.6187,
+      "step": 2551
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.3730870410309311e-05,
+      "loss": 1.4224,
+      "step": 2552
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.3716085375413498e-05,
+      "loss": 1.5059,
+      "step": 2553
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.3701304149797898e-05,
+      "loss": 1.5843,
+      "step": 2554
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.368652674242283e-05,
+      "loss": 1.531,
+      "step": 2555
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.3671753162246291e-05,
+      "loss": 1.8052,
+      "step": 2556
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3656983418223953e-05,
+      "loss": 1.9043,
+      "step": 2557
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3642217519309162e-05,
+      "loss": 1.5591,
+      "step": 2558
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3627455474452944e-05,
+      "loss": 1.5402,
+      "step": 2559
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3612697292603982e-05,
+      "loss": 1.5502,
+      "step": 2560
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3597942982708619e-05,
+      "loss": 1.4424,
+      "step": 2561
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.358319255371085e-05,
+      "loss": 1.5025,
+      "step": 2562
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3568446014552317e-05,
+      "loss": 1.3678,
+      "step": 2563
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3553703374172305e-05,
+      "loss": 1.3396,
+      "step": 2564
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3538964641507733e-05,
+      "loss": 1.2705,
+      "step": 2565
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3524229825493151e-05,
+      "loss": 1.4648,
+      "step": 2566
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3509498935060746e-05,
+      "loss": 1.7544,
+      "step": 2567
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.349477197914031e-05,
+      "loss": 1.6816,
+      "step": 2568
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3480048966659252e-05,
+      "loss": 1.4727,
+      "step": 2569
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.346532990654261e-05,
+      "loss": 1.6932,
+      "step": 2570
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3450614807712991e-05,
+      "loss": 1.5797,
+      "step": 2571
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3435903679090632e-05,
+      "loss": 1.2219,
+      "step": 2572
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3421196529593346e-05,
+      "loss": 1.3347,
+      "step": 2573
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3406493368136543e-05,
+      "loss": 1.5122,
+      "step": 2574
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3391794203633212e-05,
+      "loss": 1.4627,
+      "step": 2575
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.337709904499392e-05,
+      "loss": 1.4572,
+      "step": 2576
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3362407901126804e-05,
+      "loss": 1.6636,
+      "step": 2577
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3347720780937566e-05,
+      "loss": 1.5064,
+      "step": 2578
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3333037693329472e-05,
+      "loss": 1.5121,
+      "step": 2579
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.331835864720334e-05,
+      "loss": 1.4914,
+      "step": 2580
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.330368365145755e-05,
+      "loss": 1.1438,
+      "step": 2581
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3289012714988006e-05,
+      "loss": 1.5583,
+      "step": 2582
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3274345846688172e-05,
+      "loss": 1.3027,
+      "step": 2583
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3259683055449024e-05,
+      "loss": 1.5185,
+      "step": 2584
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3245024350159085e-05,
+      "loss": 1.4065,
+      "step": 2585
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3230369739704401e-05,
+      "loss": 1.9622,
+      "step": 2586
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3215719232968521e-05,
+      "loss": 1.2722,
+      "step": 2587
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3201072838832518e-05,
+      "loss": 1.5072,
+      "step": 2588
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.318643056617497e-05,
+      "loss": 1.2628,
+      "step": 2589
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3171792423871953e-05,
+      "loss": 1.5758,
+      "step": 2590
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3157158420797042e-05,
+      "loss": 1.4549,
+      "step": 2591
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3142528565821307e-05,
+      "loss": 1.5473,
+      "step": 2592
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3127902867813294e-05,
+      "loss": 1.454,
+      "step": 2593
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3113281335639036e-05,
+      "loss": 1.5021,
+      "step": 2594
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3098663978162036e-05,
+      "loss": 1.9391,
+      "step": 2595
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3084050804243276e-05,
+      "loss": 1.3916,
+      "step": 2596
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3069441822741195e-05,
+      "loss": 1.6378,
+      "step": 2597
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.305483704251168e-05,
+      "loss": 1.8472,
+      "step": 2598
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.3040236472408087e-05,
+      "loss": 1.6673,
+      "step": 2599
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.3025640121281215e-05,
+      "loss": 1.588,
+      "step": 2600
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.3011047997979307e-05,
+      "loss": 1.1388,
+      "step": 2601
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.299646011134804e-05,
+      "loss": 1.5879,
+      "step": 2602
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2981876470230521e-05,
+      "loss": 1.6243,
+      "step": 2603
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2967297083467295e-05,
+      "loss": 1.4987,
+      "step": 2604
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2952721959896306e-05,
+      "loss": 1.4479,
+      "step": 2605
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.293815110835294e-05,
+      "loss": 1.567,
+      "step": 2606
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2923584537669971e-05,
+      "loss": 1.4639,
+      "step": 2607
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2909022256677593e-05,
+      "loss": 1.8277,
+      "step": 2608
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2894464274203393e-05,
+      "loss": 1.9616,
+      "step": 2609
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.287991059907236e-05,
+      "loss": 1.4849,
+      "step": 2610
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2865361240106848e-05,
+      "loss": 1.5768,
+      "step": 2611
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2850816206126623e-05,
+      "loss": 1.5869,
+      "step": 2612
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2836275505948815e-05,
+      "loss": 1.454,
+      "step": 2613
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.282173914838793e-05,
+      "loss": 1.47,
+      "step": 2614
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2807207142255842e-05,
+      "loss": 1.5097,
+      "step": 2615
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2792679496361786e-05,
+      "loss": 1.4619,
+      "step": 2616
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2778156219512356e-05,
+      "loss": 1.4948,
+      "step": 2617
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2763637320511494e-05,
+      "loss": 1.5264,
+      "step": 2618
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2749122808160494e-05,
+      "loss": 1.5691,
+      "step": 2619
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2734612691257979e-05,
+      "loss": 1.6708,
+      "step": 2620
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.272010697859992e-05,
+      "loss": 1.4545,
+      "step": 2621
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2705605678979613e-05,
+      "loss": 1.4839,
+      "step": 2622
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2691108801187682e-05,
+      "loss": 1.2014,
+      "step": 2623
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2676616354012066e-05,
+      "loss": 1.2213,
+      "step": 2624
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2662128346238016e-05,
+      "loss": 1.3598,
+      "step": 2625
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2647644786648102e-05,
+      "loss": 1.689,
+      "step": 2626
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2633165684022184e-05,
+      "loss": 1.4656,
+      "step": 2627
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2618691047137435e-05,
+      "loss": 1.4744,
+      "step": 2628
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2604220884768307e-05,
+      "loss": 1.5898,
+      "step": 2629
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2589755205686552e-05,
+      "loss": 1.7249,
+      "step": 2630
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2575294018661195e-05,
+      "loss": 1.6533,
+      "step": 2631
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2560837332458539e-05,
+      "loss": 1.4398,
+      "step": 2632
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2546385155842166e-05,
+      "loss": 1.5619,
+      "step": 2633
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2531937497572919e-05,
+      "loss": 1.2959,
+      "step": 2634
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2517494366408898e-05,
+      "loss": 2.1649,
+      "step": 2635
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2503055771105465e-05,
+      "loss": 1.2305,
+      "step": 2636
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2488621720415239e-05,
+      "loss": 1.7645,
+      "step": 2637
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.247419222308806e-05,
+      "loss": 1.2403,
+      "step": 2638
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2459767287871025e-05,
+      "loss": 1.6302,
+      "step": 2639
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2445346923508472e-05,
+      "loss": 1.5173,
+      "step": 2640
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.2430931138741955e-05,
+      "loss": 1.307,
+      "step": 2641
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.2416519942310256e-05,
+      "loss": 1.3566,
+      "step": 2642
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.240211334294938e-05,
+      "loss": 1.4845,
+      "step": 2643
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.238771134939254e-05,
+      "loss": 1.5717,
+      "step": 2644
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.2373313970370156e-05,
+      "loss": 1.4918,
+      "step": 2645
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.2358921214609851e-05,
+      "loss": 1.6705,
+      "step": 2646
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.2344533090836457e-05,
+      "loss": 1.2474,
+      "step": 2647
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.2330149607771979e-05,
+      "loss": 1.8479,
+      "step": 2648
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.2315770774135624e-05,
+      "loss": 1.4977,
+      "step": 2649
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.2301396598643777e-05,
+      "loss": 1.7927,
+      "step": 2650
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.2287027090009992e-05,
+      "loss": 1.8214,
+      "step": 2651
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.2272662256944998e-05,
+      "loss": 1.1766,
+      "step": 2652
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.2258302108156691e-05,
+      "loss": 1.3477,
+      "step": 2653
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.2243946652350136e-05,
+      "loss": 1.6555,
+      "step": 2654
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.2229595898227532e-05,
+      "loss": 1.4378,
+      "step": 2655
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.2215249854488249e-05,
+      "loss": 1.697,
+      "step": 2656
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.2200908529828794e-05,
+      "loss": 1.6527,
+      "step": 2657
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.2186571932942805e-05,
+      "loss": 1.3804,
+      "step": 2658
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.2172240072521068e-05,
+      "loss": 1.6465,
+      "step": 2659
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.2157912957251485e-05,
+      "loss": 1.5726,
+      "step": 2660
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.2143590595819092e-05,
+      "loss": 1.4793,
+      "step": 2661
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.2129272996906036e-05,
+      "loss": 1.7125,
+      "step": 2662
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.2114960169191581e-05,
+      "loss": 1.6656,
+      "step": 2663
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.2100652121352106e-05,
+      "loss": 1.5352,
+      "step": 2664
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.2086348862061068e-05,
+      "loss": 2.0425,
+      "step": 2665
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.2072050399989043e-05,
+      "loss": 1.7284,
+      "step": 2666
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.20577567438037e-05,
+      "loss": 1.4878,
+      "step": 2667
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.2043467902169778e-05,
+      "loss": 1.4592,
+      "step": 2668
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.2029183883749113e-05,
+      "loss": 1.1302,
+      "step": 2669
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.2014904697200612e-05,
+      "loss": 1.4971,
+      "step": 2670
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.2000630351180247e-05,
+      "loss": 1.1075,
+      "step": 2671
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1986360854341068e-05,
+      "loss": 1.6916,
+      "step": 2672
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1972096215333176e-05,
+      "loss": 1.4558,
+      "step": 2673
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1957836442803727e-05,
+      "loss": 1.2084,
+      "step": 2674
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1943581545396934e-05,
+      "loss": 1.8218,
+      "step": 2675
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1929331531754044e-05,
+      "loss": 1.7064,
+      "step": 2676
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.191508641051336e-05,
+      "loss": 1.5698,
+      "step": 2677
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1900846190310204e-05,
+      "loss": 1.7531,
+      "step": 2678
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.188661087977693e-05,
+      "loss": 1.8175,
+      "step": 2679
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1872380487542917e-05,
+      "loss": 1.5177,
+      "step": 2680
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1858155022234565e-05,
+      "loss": 1.3862,
+      "step": 2681
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1843934492475286e-05,
+      "loss": 1.4711,
+      "step": 2682
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.1829718906885504e-05,
+      "loss": 1.3359,
+      "step": 2683
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.1815508274082634e-05,
+      "loss": 1.8342,
+      "step": 2684
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.1801302602681103e-05,
+      "loss": 1.3898,
+      "step": 2685
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.1787101901292321e-05,
+      "loss": 1.4031,
+      "step": 2686
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.1772906178524691e-05,
+      "loss": 1.4814,
+      "step": 2687
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.1758715442983592e-05,
+      "loss": 2.2663,
+      "step": 2688
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.1744529703271383e-05,
+      "loss": 1.4726,
+      "step": 2689
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.1730348967987394e-05,
+      "loss": 1.5892,
+      "step": 2690
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.1716173245727927e-05,
+      "loss": 1.3422,
+      "step": 2691
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.1702002545086228e-05,
+      "loss": 1.5112,
+      "step": 2692
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.168783687465252e-05,
+      "loss": 1.6255,
+      "step": 2693
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.1673676243013965e-05,
+      "loss": 1.2162,
+      "step": 2694
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.1659520658754673e-05,
+      "loss": 1.6152,
+      "step": 2695
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.1645370130455698e-05,
+      "loss": 2.01,
+      "step": 2696
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.1631224666695019e-05,
+      "loss": 1.6486,
+      "step": 2697
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.1617084276047552e-05,
+      "loss": 1.6569,
+      "step": 2698
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.1602948967085141e-05,
+      "loss": 1.4678,
+      "step": 2699
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.1588818748376544e-05,
+      "loss": 1.7867,
+      "step": 2700
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.1574693628487434e-05,
+      "loss": 1.8689,
+      "step": 2701
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.1560573615980393e-05,
+      "loss": 1.8724,
+      "step": 2702
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.1546458719414917e-05,
+      "loss": 1.9627,
+      "step": 2703
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.1532348947347375e-05,
+      "loss": 1.422,
+      "step": 2704
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.1518244308331056e-05,
+      "loss": 1.3565,
+      "step": 2705
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.1504144810916127e-05,
+      "loss": 1.4577,
+      "step": 2706
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.1490050463649635e-05,
+      "loss": 1.2424,
+      "step": 2707
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.1475961275075513e-05,
+      "loss": 1.5016,
+      "step": 2708
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.1461877253734562e-05,
+      "loss": 1.3205,
+      "step": 2709
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.1447798408164447e-05,
+      "loss": 1.8553,
+      "step": 2710
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.1433724746899713e-05,
+      "loss": 1.4334,
+      "step": 2711
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.1419656278471729e-05,
+      "loss": 1.4359,
+      "step": 2712
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.1405593011408745e-05,
+      "loss": 1.4337,
+      "step": 2713
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.139153495423585e-05,
+      "loss": 1.717,
+      "step": 2714
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.1377482115474972e-05,
+      "loss": 1.5188,
+      "step": 2715
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.136343450364488e-05,
+      "loss": 1.6811,
+      "step": 2716
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.1349392127261169e-05,
+      "loss": 1.3939,
+      "step": 2717
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.1335354994836276e-05,
+      "loss": 1.5015,
+      "step": 2718
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.1321323114879424e-05,
+      "loss": 1.7598,
+      "step": 2719
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.1307296495896684e-05,
+      "loss": 1.4953,
+      "step": 2720
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.1293275146390931e-05,
+      "loss": 1.2761,
+      "step": 2721
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.1279259074861841e-05,
+      "loss": 1.6056,
+      "step": 2722
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.1265248289805892e-05,
+      "loss": 1.5955,
+      "step": 2723
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.1251242799716357e-05,
+      "loss": 1.63,
+      "step": 2724
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 1.12372426130833e-05,
+      "loss": 1.4696,
+      "step": 2725
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 1.1223247738393575e-05,
+      "loss": 1.6774,
+      "step": 2726
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 1.1209258184130802e-05,
+      "loss": 1.9409,
+      "step": 2727
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 1.1195273958775391e-05,
+      "loss": 1.5044,
+      "step": 2728
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 1.118129507080451e-05,
+      "loss": 1.7858,
+      "step": 2729
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 1.11673215286921e-05,
+      "loss": 1.1876,
+      "step": 2730
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 1.115335334090886e-05,
+      "loss": 1.8247,
+      "step": 2731
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 1.113939051592225e-05,
+      "loss": 1.417,
+      "step": 2732
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 1.1125433062196449e-05,
+      "loss": 1.5365,
+      "step": 2733
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 1.1111480988192413e-05,
+      "loss": 1.417,
+      "step": 2734
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 1.1097534302367824e-05,
+      "loss": 1.481,
+      "step": 2735
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 1.1083593013177101e-05,
+      "loss": 1.4824,
+      "step": 2736
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 1.1069657129071388e-05,
+      "loss": 1.7476,
+      "step": 2737
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 1.1055726658498554e-05,
+      "loss": 1.3819,
+      "step": 2738
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 1.1041801609903185e-05,
+      "loss": 1.6339,
+      "step": 2739
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 1.1027881991726583e-05,
+      "loss": 1.4118,
+      "step": 2740
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 1.1013967812406758e-05,
+      "loss": 1.5586,
+      "step": 2741
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 1.100005908037842e-05,
+      "loss": 1.8291,
+      "step": 2742
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 1.098615580407298e-05,
+      "loss": 1.5247,
+      "step": 2743
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 1.0972257991918535e-05,
+      "loss": 1.8677,
+      "step": 2744
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 1.0958365652339884e-05,
+      "loss": 1.8116,
+      "step": 2745
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 1.0944478793758502e-05,
+      "loss": 1.738,
+      "step": 2746
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 1.0930597424592522e-05,
+      "loss": 1.7752,
+      "step": 2747
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 1.0916721553256777e-05,
+      "loss": 1.6519,
+      "step": 2748
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 1.0902851188162755e-05,
+      "loss": 1.4888,
+      "step": 2749
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 1.0888986337718608e-05,
+      "loss": 1.5854,
+      "step": 2750
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 1.0875127010329147e-05,
+      "loss": 1.4487,
+      "step": 2751
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 1.086127321439583e-05,
+      "loss": 1.3848,
+      "step": 2752
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 1.0847424958316764e-05,
+      "loss": 1.4186,
+      "step": 2753
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 1.08335822504867e-05,
+      "loss": 1.6957,
+      "step": 2754
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 1.0819745099297022e-05,
+      "loss": 1.8438,
+      "step": 2755
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 1.0805913513135751e-05,
+      "loss": 1.4854,
+      "step": 2756
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 1.079208750038753e-05,
+      "loss": 1.4689,
+      "step": 2757
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 1.0778267069433621e-05,
+      "loss": 1.5201,
+      "step": 2758
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 1.0764452228651918e-05,
+      "loss": 1.7206,
+      "step": 2759
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 1.0750642986416894e-05,
+      "loss": 1.9609,
+      "step": 2760
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 1.0736839351099663e-05,
+      "loss": 1.4789,
+      "step": 2761
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 1.0723041331067917e-05,
+      "loss": 1.7442,
+      "step": 2762
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 1.070924893468596e-05,
+      "loss": 1.4554,
+      "step": 2763
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 1.0695462170314677e-05,
+      "loss": 1.2415,
+      "step": 2764
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 1.0681681046311538e-05,
+      "loss": 1.7597,
+      "step": 2765
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 1.0667905571030599e-05,
+      "loss": 1.6345,
+      "step": 2766
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 1.0654135752822495e-05,
+      "loss": 1.7483,
+      "step": 2767
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 1.0640371600034416e-05,
+      "loss": 1.2986,
+      "step": 2768
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 1.0626613121010139e-05,
+      "loss": 1.3781,
+      "step": 2769
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 1.0612860324089985e-05,
+      "loss": 2.0882,
+      "step": 2770
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 1.0599113217610837e-05,
+      "loss": 1.4265,
+      "step": 2771
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 1.0585371809906127e-05,
+      "loss": 1.5648,
+      "step": 2772
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 1.0571636109305845e-05,
+      "loss": 1.4281,
+      "step": 2773
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 1.0557906124136493e-05,
+      "loss": 1.2389,
+      "step": 2774
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 1.0544181862721129e-05,
+      "loss": 1.5143,
+      "step": 2775
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 1.053046333337934e-05,
+      "loss": 1.6062,
+      "step": 2776
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 1.0516750544427236e-05,
+      "loss": 1.3932,
+      "step": 2777
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 1.0503043504177446e-05,
+      "loss": 1.6872,
+      "step": 2778
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 1.0489342220939113e-05,
+      "loss": 1.6824,
+      "step": 2779
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 1.0475646703017906e-05,
+      "loss": 1.5817,
+      "step": 2780
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 1.0461956958715965e-05,
+      "loss": 1.3772,
+      "step": 2781
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 1.0448272996331957e-05,
+      "loss": 1.4509,
+      "step": 2782
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 1.043459482416104e-05,
+      "loss": 1.4362,
+      "step": 2783
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 1.0420922450494855e-05,
+      "loss": 1.4654,
+      "step": 2784
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 1.0407255883621536e-05,
+      "loss": 1.6129,
+      "step": 2785
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 1.0393595131825695e-05,
+      "loss": 1.7833,
+      "step": 2786
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 1.037994020338842e-05,
+      "loss": 1.5477,
+      "step": 2787
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 1.0366291106587251e-05,
+      "loss": 1.6444,
+      "step": 2788
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 1.035264784969622e-05,
+      "loss": 1.8021,
+      "step": 2789
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 1.0339010440985801e-05,
+      "loss": 1.7615,
+      "step": 2790
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 1.0325378888722937e-05,
+      "loss": 1.433,
+      "step": 2791
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 1.0311753201171002e-05,
+      "loss": 1.6194,
+      "step": 2792
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 1.0298133386589831e-05,
+      "loss": 1.6553,
+      "step": 2793
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 1.0284519453235693e-05,
+      "loss": 1.4193,
+      "step": 2794
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 1.0270911409361291e-05,
+      "loss": 1.9524,
+      "step": 2795
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 1.0257309263215758e-05,
+      "loss": 1.4939,
+      "step": 2796
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 1.0243713023044655e-05,
+      "loss": 1.3668,
+      "step": 2797
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 1.023012269708996e-05,
+      "loss": 1.4401,
+      "step": 2798
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 1.0216538293590061e-05,
+      "loss": 1.4632,
+      "step": 2799
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 1.0202959820779776e-05,
+      "loss": 1.5697,
+      "step": 2800
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 1.0189387286890295e-05,
+      "loss": 1.5604,
+      "step": 2801
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 1.0175820700149233e-05,
+      "loss": 1.311,
+      "step": 2802
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 1.0162260068780595e-05,
+      "loss": 1.3021,
+      "step": 2803
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 1.0148705401004766e-05,
+      "loss": 1.633,
+      "step": 2804
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 1.0135156705038533e-05,
+      "loss": 1.5187,
+      "step": 2805
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 1.0121613989095047e-05,
+      "loss": 1.1938,
+      "step": 2806
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 1.0108077261383843e-05,
+      "loss": 1.8009,
+      "step": 2807
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 1.009454653011082e-05,
+      "loss": 1.5073,
+      "step": 2808
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 1.0081021803478252e-05,
+      "loss": 1.6606,
+      "step": 2809
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 1.006750308968476e-05,
+      "loss": 1.8056,
+      "step": 2810
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 1.0053990396925328e-05,
+      "loss": 1.3724,
+      "step": 2811
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 1.0040483733391289e-05,
+      "loss": 1.6012,
+      "step": 2812
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 1.0026983107270327e-05,
+      "loss": 1.4344,
+      "step": 2813
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 1.0013488526746444e-05,
+      "loss": 1.3578,
+      "step": 2814
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 1.3365,
+      "step": 2815
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 9.986517535207686e-06,
+      "loss": 1.6094,
+      "step": 2816
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 9.973041140542502e-06,
+      "loss": 1.3857,
+      "step": 2817
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 9.959570824173776e-06,
+      "loss": 1.733,
+      "step": 2818
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 9.946106594267155e-06,
+      "loss": 1.3319,
+      "step": 2819
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 9.932648458984594e-06,
+      "loss": 1.4336,
+      "step": 2820
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 9.919196426484352e-06,
+      "loss": 1.663,
+      "step": 2821
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 9.905750504920988e-06,
+      "loss": 1.2911,
+      "step": 2822
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 9.892310702445357e-06,
+      "loss": 1.6328,
+      "step": 2823
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 9.878877027204611e-06,
+      "loss": 1.2483,
+      "step": 2824
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 9.865449487342178e-06,
+      "loss": 1.4838,
+      "step": 2825
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 9.852028090997773e-06,
+      "loss": 1.3626,
+      "step": 2826
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 9.83861284630739e-06,
+      "loss": 1.6142,
+      "step": 2827
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 9.825203761403275e-06,
+      "loss": 1.7158,
+      "step": 2828
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 9.811800844413964e-06,
+      "loss": 1.7274,
+      "step": 2829
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 9.798404103464238e-06,
+      "loss": 1.1903,
+      "step": 2830
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 9.785013546675145e-06,
+      "loss": 1.514,
+      "step": 2831
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 9.771629182163975e-06,
+      "loss": 1.8397,
+      "step": 2832
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 9.758251018044274e-06,
+      "loss": 1.6419,
+      "step": 2833
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 9.744879062425818e-06,
+      "loss": 1.3573,
+      "step": 2834
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 9.731513323414627e-06,
+      "loss": 1.6292,
+      "step": 2835
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 9.718153809112954e-06,
+      "loss": 1.4934,
+      "step": 2836
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 9.704800527619271e-06,
+      "loss": 2.0658,
+      "step": 2837
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 9.691453487028278e-06,
+      "loss": 1.6922,
+      "step": 2838
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 9.678112695430888e-06,
+      "loss": 1.5602,
+      "step": 2839
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 9.664778160914236e-06,
+      "loss": 1.4883,
+      "step": 2840
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 9.651449891561642e-06,
+      "loss": 2.0361,
+      "step": 2841
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 9.63812789545265e-06,
+      "loss": 1.6767,
+      "step": 2842
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 9.624812180662987e-06,
+      "loss": 1.7158,
+      "step": 2843
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 9.611502755264586e-06,
+      "loss": 1.295,
+      "step": 2844
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 9.598199627325555e-06,
+      "loss": 1.6996,
+      "step": 2845
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 9.58490280491019e-06,
+      "loss": 1.317,
+      "step": 2846
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 9.571612296078962e-06,
+      "loss": 1.9413,
+      "step": 2847
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 9.558328108888521e-06,
+      "loss": 1.293,
+      "step": 2848
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 9.545050251391685e-06,
+      "loss": 1.3997,
+      "step": 2849
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 9.531778731637416e-06,
+      "loss": 1.279,
+      "step": 2850
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 9.518513557670852e-06,
+      "loss": 1.6311,
+      "step": 2851
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 9.505254737533288e-06,
+      "loss": 1.8417,
+      "step": 2852
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 9.492002279262154e-06,
+      "loss": 1.8815,
+      "step": 2853
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 9.478756190891032e-06,
+      "loss": 1.6442,
+      "step": 2854
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 9.465516480449637e-06,
+      "loss": 1.8327,
+      "step": 2855
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 9.452283155963832e-06,
+      "loss": 1.4884,
+      "step": 2856
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 9.439056225455583e-06,
+      "loss": 1.531,
+      "step": 2857
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 9.425835696943e-06,
+      "loss": 1.326,
+      "step": 2858
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 9.412621578440303e-06,
+      "loss": 1.506,
+      "step": 2859
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 9.39941387795784e-06,
+      "loss": 1.3937,
+      "step": 2860
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 9.386212603502047e-06,
+      "loss": 1.3287,
+      "step": 2861
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 9.373017763075485e-06,
+      "loss": 1.8481,
+      "step": 2862
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 9.3598293646768e-06,
+      "loss": 1.7084,
+      "step": 2863
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 9.346647416300742e-06,
+      "loss": 1.912,
+      "step": 2864
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 9.333471925938146e-06,
+      "loss": 1.4633,
+      "step": 2865
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 9.320302901575935e-06,
+      "loss": 1.4562,
+      "step": 2866
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 9.30714035119712e-06,
+      "loss": 1.6216,
+      "step": 2867
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 9.293984282780758e-06,
+      "loss": 1.2655,
+      "step": 2868
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 9.28083470430201e-06,
+      "loss": 1.2996,
+      "step": 2869
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 9.267691623732087e-06,
+      "loss": 1.4443,
+      "step": 2870
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 9.254555049038266e-06,
+      "loss": 1.9278,
+      "step": 2871
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 9.241424988183878e-06,
+      "loss": 1.7681,
+      "step": 2872
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 9.228301449128308e-06,
+      "loss": 1.9479,
+      "step": 2873
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 9.215184439826981e-06,
+      "loss": 1.5484,
+      "step": 2874
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 9.202073968231369e-06,
+      "loss": 1.4043,
+      "step": 2875
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 9.18897004228898e-06,
+      "loss": 1.7652,
+      "step": 2876
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 9.175872669943355e-06,
+      "loss": 1.6967,
+      "step": 2877
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 9.162781859134061e-06,
+      "loss": 1.6073,
+      "step": 2878
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 9.149697617796687e-06,
+      "loss": 1.6989,
+      "step": 2879
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 9.136619953862838e-06,
+      "loss": 1.5986,
+      "step": 2880
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 9.123548875260144e-06,
+      "loss": 1.3991,
+      "step": 2881
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 9.110484389912218e-06,
+      "loss": 1.6632,
+      "step": 2882
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 9.097426505738695e-06,
+      "loss": 1.0419,
+      "step": 2883
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 9.084375230655209e-06,
+      "loss": 1.2821,
+      "step": 2884
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 9.071330572573382e-06,
+      "loss": 1.4647,
+      "step": 2885
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 9.058292539400824e-06,
+      "loss": 1.8963,
+      "step": 2886
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 9.04526113904113e-06,
+      "loss": 1.3033,
+      "step": 2887
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 9.03223637939388e-06,
+      "loss": 1.8094,
+      "step": 2888
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 9.01921826835462e-06,
+      "loss": 1.4723,
+      "step": 2889
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 9.006206813814868e-06,
+      "loss": 1.3087,
+      "step": 2890
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 8.993202023662114e-06,
+      "loss": 1.2881,
+      "step": 2891
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 8.980203905779794e-06,
+      "loss": 1.5642,
+      "step": 2892
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 8.967212468047312e-06,
+      "loss": 1.4341,
+      "step": 2893
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 8.954227718340025e-06,
+      "loss": 1.5455,
+      "step": 2894
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 8.941249664529215e-06,
+      "loss": 1.4381,
+      "step": 2895
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 8.928278314482124e-06,
+      "loss": 1.8877,
+      "step": 2896
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 8.915313676061925e-06,
+      "loss": 1.3971,
+      "step": 2897
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 8.902355757127723e-06,
+      "loss": 1.7332,
+      "step": 2898
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 8.88940456553455e-06,
+      "loss": 1.062,
+      "step": 2899
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 8.876460109133356e-06,
+      "loss": 1.6749,
+      "step": 2900
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 8.863522395771014e-06,
+      "loss": 1.2884,
+      "step": 2901
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 8.850591433290305e-06,
+      "loss": 1.7454,
+      "step": 2902
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 8.83766722952992e-06,
+      "loss": 1.4954,
+      "step": 2903
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 8.824749792324452e-06,
+      "loss": 1.5715,
+      "step": 2904
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 8.811839129504391e-06,
+      "loss": 1.8461,
+      "step": 2905
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 8.798935248896126e-06,
+      "loss": 1.2963,
+      "step": 2906
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 8.78603815832192e-06,
+      "loss": 1.753,
+      "step": 2907
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 8.77314786559995e-06,
+      "loss": 1.4852,
+      "step": 2908
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 8.76026437854423e-06,
+      "loss": 1.259,
+      "step": 2909
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 8.747387704964678e-06,
+      "loss": 1.2822,
+      "step": 2910
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 8.734517852667077e-06,
+      "loss": 1.5785,
+      "step": 2911
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 8.721654829453072e-06,
+      "loss": 1.4628,
+      "step": 2912
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 8.708798643120172e-06,
+      "loss": 1.335,
+      "step": 2913
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 8.695949301461728e-06,
+      "loss": 1.7141,
+      "step": 2914
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 8.68310681226696e-06,
+      "loss": 1.613,
+      "step": 2915
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 8.670271183320928e-06,
+      "loss": 1.1629,
+      "step": 2916
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 8.657442422404527e-06,
+      "loss": 1.5205,
+      "step": 2917
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 8.6446205372945e-06,
+      "loss": 1.113,
+      "step": 2918
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 8.631805535763402e-06,
+      "loss": 1.4083,
+      "step": 2919
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 8.618997425579634e-06,
+      "loss": 1.8069,
+      "step": 2920
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 8.606196214507416e-06,
+      "loss": 1.4242,
+      "step": 2921
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 8.59340191030678e-06,
+      "loss": 1.7203,
+      "step": 2922
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 8.58061452073358e-06,
+      "loss": 1.766,
+      "step": 2923
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 8.567834053539466e-06,
+      "loss": 1.4567,
+      "step": 2924
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 8.55506051647191e-06,
+      "loss": 1.5017,
+      "step": 2925
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 8.542293917274158e-06,
+      "loss": 1.4074,
+      "step": 2926
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 8.529534263685268e-06,
+      "loss": 1.2303,
+      "step": 2927
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 8.516781563440085e-06,
+      "loss": 1.5954,
+      "step": 2928
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 8.504035824269236e-06,
+      "loss": 1.5022,
+      "step": 2929
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 8.491297053899135e-06,
+      "loss": 1.8951,
+      "step": 2930
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 8.478565260051959e-06,
+      "loss": 1.4753,
+      "step": 2931
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 8.465840450445666e-06,
+      "loss": 1.4448,
+      "step": 2932
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 8.453122632793977e-06,
+      "loss": 1.6566,
+      "step": 2933
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 8.440411814806377e-06,
+      "loss": 1.1525,
+      "step": 2934
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 8.427708004188111e-06,
+      "loss": 1.372,
+      "step": 2935
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 8.415011208640154e-06,
+      "loss": 1.3116,
+      "step": 2936
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 8.402321435859255e-06,
+      "loss": 1.6859,
+      "step": 2937
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 8.389638693537896e-06,
+      "loss": 1.3074,
+      "step": 2938
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 8.376962989364297e-06,
+      "loss": 1.8422,
+      "step": 2939
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 8.364294331022411e-06,
+      "loss": 1.8411,
+      "step": 2940
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 8.351632726191921e-06,
+      "loss": 1.4699,
+      "step": 2941
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 8.338978182548234e-06,
+      "loss": 1.7167,
+      "step": 2942
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 8.326330707762476e-06,
+      "loss": 1.1935,
+      "step": 2943
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 8.31369030950149e-06,
+      "loss": 1.7694,
+      "step": 2944
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 8.301056995427823e-06,
+      "loss": 1.4604,
+      "step": 2945
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 8.288430773199735e-06,
+      "loss": 1.9495,
+      "step": 2946
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 8.275811650471186e-06,
+      "loss": 1.6463,
+      "step": 2947
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 8.263199634891826e-06,
+      "loss": 1.3826,
+      "step": 2948
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 8.250594734107012e-06,
+      "loss": 1.6183,
+      "step": 2949
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 8.237996955757758e-06,
+      "loss": 1.8059,
+      "step": 2950
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 8.225406307480792e-06,
+      "loss": 1.5125,
+      "step": 2951
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 8.212822796908504e-06,
+      "loss": 1.647,
+      "step": 2952
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 8.200246431668961e-06,
+      "loss": 1.5127,
+      "step": 2953
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 8.187677219385899e-06,
+      "loss": 1.1933,
+      "step": 2954
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 8.175115167678715e-06,
+      "loss": 1.5523,
+      "step": 2955
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 8.162560284162468e-06,
+      "loss": 1.6955,
+      "step": 2956
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 8.150012576447872e-06,
+      "loss": 1.2973,
+      "step": 2957
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 8.137472052141286e-06,
+      "loss": 1.4322,
+      "step": 2958
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 8.124938718844724e-06,
+      "loss": 2.0217,
+      "step": 2959
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 8.112412584155832e-06,
+      "loss": 1.2762,
+      "step": 2960
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 8.099893655667892e-06,
+      "loss": 1.5377,
+      "step": 2961
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 8.087381940969834e-06,
+      "loss": 1.6437,
+      "step": 2962
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 8.074877447646186e-06,
+      "loss": 1.531,
+      "step": 2963
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 8.06238018327712e-06,
+      "loss": 1.7565,
+      "step": 2964
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 8.04989015543842e-06,
+      "loss": 1.6079,
+      "step": 2965
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 8.037407371701487e-06,
+      "loss": 1.7219,
+      "step": 2966
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 8.024931839633323e-06,
+      "loss": 1.4672,
+      "step": 2967
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 8.012463566796542e-06,
+      "loss": 1.9537,
+      "step": 2968
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 8.000002560749352e-06,
+      "loss": 1.5811,
+      "step": 2969
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 7.987548829045558e-06,
+      "loss": 1.43,
+      "step": 2970
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 7.975102379234556e-06,
+      "loss": 1.4324,
+      "step": 2971
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 7.962663218861324e-06,
+      "loss": 1.2979,
+      "step": 2972
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 7.950231355466426e-06,
+      "loss": 1.9095,
+      "step": 2973
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 7.937806796586001e-06,
+      "loss": 1.6857,
+      "step": 2974
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 7.92538954975176e-06,
+      "loss": 1.6272,
+      "step": 2975
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 7.912979622490984e-06,
+      "loss": 1.2397,
+      "step": 2976
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 7.900577022326503e-06,
+      "loss": 1.8098,
+      "step": 2977
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 7.888181756776725e-06,
+      "loss": 1.5566,
+      "step": 2978
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 7.8757938333556e-06,
+      "loss": 1.8447,
+      "step": 2979
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 7.863413259572634e-06,
+      "loss": 1.7123,
+      "step": 2980
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 7.851040042932869e-06,
+      "loss": 1.3322,
+      "step": 2981
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 7.838674190936897e-06,
+      "loss": 1.91,
+      "step": 2982
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 7.826315711080838e-06,
+      "loss": 1.5103,
+      "step": 2983
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 7.813964610856346e-06,
+      "loss": 1.2631,
+      "step": 2984
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 7.801620897750602e-06,
+      "loss": 1.3333,
+      "step": 2985
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 7.789284579246308e-06,
+      "loss": 1.5915,
+      "step": 2986
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 7.77695566282169e-06,
+      "loss": 1.4337,
+      "step": 2987
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 7.764634155950466e-06,
+      "loss": 1.4089,
+      "step": 2988
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 7.752320066101889e-06,
+      "loss": 1.5998,
+      "step": 2989
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 7.740013400740695e-06,
+      "loss": 1.7803,
+      "step": 2990
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 7.727714167327136e-06,
+      "loss": 1.5089,
+      "step": 2991
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 7.715422373316948e-06,
+      "loss": 1.4767,
+      "step": 2992
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 7.70313802616136e-06,
+      "loss": 1.5976,
+      "step": 2993
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 7.690861133307093e-06,
+      "loss": 1.8934,
+      "step": 2994
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 7.678591702196329e-06,
+      "loss": 1.8553,
+      "step": 2995
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 7.66632974026675e-06,
+      "loss": 1.5636,
+      "step": 2996
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 7.654075254951501e-06,
+      "loss": 1.4064,
+      "step": 2997
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 7.641828253679194e-06,
+      "loss": 1.3159,
+      "step": 2998
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 7.629588743873903e-06,
+      "loss": 1.849,
+      "step": 2999
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 7.6173567329551655e-06,
+      "loss": 1.3247,
+      "step": 3000
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 7.605132228337968e-06,
+      "loss": 1.6869,
+      "step": 3001
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 7.5929152374327515e-06,
+      "loss": 1.7235,
+      "step": 3002
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 7.580705767645404e-06,
+      "loss": 1.6966,
+      "step": 3003
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 7.56850382637724e-06,
+      "loss": 1.2246,
+      "step": 3004
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 7.556309421025026e-06,
+      "loss": 1.6411,
+      "step": 3005
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 7.544122558980951e-06,
+      "loss": 1.3714,
+      "step": 3006
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 7.5319432476326384e-06,
+      "loss": 1.3944,
+      "step": 3007
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 7.51977149436313e-06,
+      "loss": 1.5817,
+      "step": 3008
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 7.507607306550888e-06,
+      "loss": 1.7454,
+      "step": 3009
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 7.495450691569786e-06,
+      "loss": 1.3643,
+      "step": 3010
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 7.483301656789108e-06,
+      "loss": 1.6135,
+      "step": 3011
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 7.471160209573543e-06,
+      "loss": 1.5535,
+      "step": 3012
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 7.459026357283181e-06,
+      "loss": 1.2622,
+      "step": 3013
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 7.446900107273509e-06,
+      "loss": 1.8498,
+      "step": 3014
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 7.4347814668954e-06,
+      "loss": 1.7284,
+      "step": 3015
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 7.42267044349513e-06,
+      "loss": 1.8606,
+      "step": 3016
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 7.41056704441433e-06,
+      "loss": 1.2496,
+      "step": 3017
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 7.3984712769900316e-06,
+      "loss": 1.5049,
+      "step": 3018
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 7.386383148554634e-06,
+      "loss": 1.3088,
+      "step": 3019
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 7.374302666435904e-06,
+      "loss": 1.465,
+      "step": 3020
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 7.362229837956975e-06,
+      "loss": 1.7779,
+      "step": 3021
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 7.350164670436344e-06,
+      "loss": 1.8081,
+      "step": 3022
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 7.338107171187854e-06,
+      "loss": 1.4781,
+      "step": 3023
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 7.326057347520707e-06,
+      "loss": 1.8476,
+      "step": 3024
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 7.314015206739456e-06,
+      "loss": 1.4018,
+      "step": 3025
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 7.301980756143985e-06,
+      "loss": 1.5169,
+      "step": 3026
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 7.2899540030295265e-06,
+      "loss": 1.9924,
+      "step": 3027
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 7.277934954686643e-06,
+      "loss": 1.4315,
+      "step": 3028
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 7.265923618401225e-06,
+      "loss": 1.4139,
+      "step": 3029
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 7.253920001454498e-06,
+      "loss": 1.4788,
+      "step": 3030
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 7.241924111122984e-06,
+      "loss": 2.0003,
+      "step": 3031
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 7.229935954678544e-06,
+      "loss": 1.4435,
+      "step": 3032
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 7.217955539388344e-06,
+      "loss": 1.7146,
+      "step": 3033
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 7.205982872514856e-06,
+      "loss": 1.4157,
+      "step": 3034
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 7.194017961315858e-06,
+      "loss": 1.4048,
+      "step": 3035
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 7.18206081304442e-06,
+      "loss": 1.1715,
+      "step": 3036
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 7.170111434948914e-06,
+      "loss": 1.9549,
+      "step": 3037
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 7.158169834272999e-06,
+      "loss": 1.5431,
+      "step": 3038
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 7.1462360182556145e-06,
+      "loss": 1.4471,
+      "step": 3039
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 7.134309994130988e-06,
+      "loss": 1.3814,
+      "step": 3040
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 7.1223917691286184e-06,
+      "loss": 1.6129,
+      "step": 3041
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 7.110481350473282e-06,
+      "loss": 1.7506,
+      "step": 3042
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 7.098578745385025e-06,
+      "loss": 1.4226,
+      "step": 3043
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 7.086683961079139e-06,
+      "loss": 1.3132,
+      "step": 3044
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 7.074797004766194e-06,
+      "loss": 1.5644,
+      "step": 3045
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 7.062917883652007e-06,
+      "loss": 1.35,
+      "step": 3046
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 7.05104660493765e-06,
+      "loss": 1.7072,
+      "step": 3047
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 7.039183175819435e-06,
+      "loss": 1.3603,
+      "step": 3048
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 7.027327603488918e-06,
+      "loss": 1.4129,
+      "step": 3049
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 7.015479895132895e-06,
+      "loss": 1.3799,
+      "step": 3050
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 7.0036400579333914e-06,
+      "loss": 1.7638,
+      "step": 3051
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 6.991808099067663e-06,
+      "loss": 1.7193,
+      "step": 3052
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 6.9799840257081865e-06,
+      "loss": 1.646,
+      "step": 3053
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 6.968167845022662e-06,
+      "loss": 1.4618,
+      "step": 3054
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 6.956359564174007e-06,
+      "loss": 1.7036,
+      "step": 3055
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 6.944559190320348e-06,
+      "loss": 1.6625,
+      "step": 3056
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 6.932766730615008e-06,
+      "loss": 1.5293,
+      "step": 3057
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 6.920982192206531e-06,
+      "loss": 1.3735,
+      "step": 3058
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 6.909205582238645e-06,
+      "loss": 1.9298,
+      "step": 3059
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 6.897436907850279e-06,
+      "loss": 1.4943,
+      "step": 3060
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 6.885676176175548e-06,
+      "loss": 1.6306,
+      "step": 3061
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 6.873923394343758e-06,
+      "loss": 1.3181,
+      "step": 3062
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 6.862178569479387e-06,
+      "loss": 1.4401,
+      "step": 3063
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 6.850441708702103e-06,
+      "loss": 1.7882,
+      "step": 3064
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 6.838712819126722e-06,
+      "loss": 1.5631,
+      "step": 3065
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 6.8269919078632515e-06,
+      "loss": 1.5322,
+      "step": 3066
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 6.815278982016855e-06,
+      "loss": 1.566,
+      "step": 3067
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 6.803574048687851e-06,
+      "loss": 1.3596,
+      "step": 3068
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 6.791877114971717e-06,
+      "loss": 1.7078,
+      "step": 3069
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 6.780188187959082e-06,
+      "loss": 1.6348,
+      "step": 3070
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 6.76850727473573e-06,
+      "loss": 1.7435,
+      "step": 3071
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 6.7568343823825555e-06,
+      "loss": 1.3811,
+      "step": 3072
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 6.745169517975625e-06,
+      "loss": 2.2434,
+      "step": 3073
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 6.733512688586126e-06,
+      "loss": 1.645,
+      "step": 3074
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 6.72186390128037e-06,
+      "loss": 1.4508,
+      "step": 3075
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 6.710223163119804e-06,
+      "loss": 1.6381,
+      "step": 3076
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 6.698590481160987e-06,
+      "loss": 1.723,
+      "step": 3077
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 6.686965862455599e-06,
+      "loss": 1.4259,
+      "step": 3078
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 6.6753493140504275e-06,
+      "loss": 1.6813,
+      "step": 3079
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 6.663740842987376e-06,
+      "loss": 1.4386,
+      "step": 3080
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 6.652140456303442e-06,
+      "loss": 1.8682,
+      "step": 3081
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 6.640548161030727e-06,
+      "loss": 1.3405,
+      "step": 3082
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 6.628963964196428e-06,
+      "loss": 1.3506,
+      "step": 3083
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 6.617387872822842e-06,
+      "loss": 1.4166,
+      "step": 3084
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 6.605819893927325e-06,
+      "loss": 1.4867,
+      "step": 3085
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 6.594260034522339e-06,
+      "loss": 1.496,
+      "step": 3086
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 6.582708301615421e-06,
+      "loss": 1.6243,
+      "step": 3087
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 6.571164702209176e-06,
+      "loss": 1.9608,
+      "step": 3088
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 6.559629243301282e-06,
+      "loss": 1.5528,
+      "step": 3089
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 6.548101931884481e-06,
+      "loss": 1.5493,
+      "step": 3090
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 6.536582774946576e-06,
+      "loss": 1.79,
+      "step": 3091
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 6.525071779470429e-06,
+      "loss": 1.6829,
+      "step": 3092
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 6.513568952433951e-06,
+      "loss": 1.4561,
+      "step": 3093
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 6.502074300810102e-06,
+      "loss": 1.6318,
+      "step": 3094
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 6.490587831566888e-06,
+      "loss": 1.5038,
+      "step": 3095
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 6.479109551667353e-06,
+      "loss": 1.6435,
+      "step": 3096
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 6.46763946806958e-06,
+      "loss": 1.3145,
+      "step": 3097
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 6.456177587726687e-06,
+      "loss": 1.6124,
+      "step": 3098
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 6.4447239175868e-06,
+      "loss": 1.7352,
+      "step": 3099
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 6.4332784645930845e-06,
+      "loss": 1.4099,
+      "step": 3100
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 6.421841235683726e-06,
+      "loss": 1.5206,
+      "step": 3101
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 6.410412237791919e-06,
+      "loss": 1.4258,
+      "step": 3102
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 6.398991477845869e-06,
+      "loss": 1.3527,
+      "step": 3103
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 6.387578962768788e-06,
+      "loss": 1.8365,
+      "step": 3104
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 6.376174699478892e-06,
+      "loss": 1.8159,
+      "step": 3105
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 6.364778694889391e-06,
+      "loss": 1.2343,
+      "step": 3106
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 6.3533909559084915e-06,
+      "loss": 1.5592,
+      "step": 3107
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 6.34201148943939e-06,
+      "loss": 1.5824,
+      "step": 3108
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 6.330640302380264e-06,
+      "loss": 1.7248,
+      "step": 3109
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 6.319277401624277e-06,
+      "loss": 1.4992,
+      "step": 3110
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 6.307922794059576e-06,
+      "loss": 1.7862,
+      "step": 3111
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 6.296576486569252e-06,
+      "loss": 1.5535,
+      "step": 3112
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.2852384860313974e-06,
+      "loss": 1.2184,
+      "step": 3113
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.273908799319055e-06,
+      "loss": 1.5397,
+      "step": 3114
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.262587433300227e-06,
+      "loss": 1.3582,
+      "step": 3115
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.251274394837874e-06,
+      "loss": 1.5799,
+      "step": 3116
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.2399696907899085e-06,
+      "loss": 1.8437,
+      "step": 3117
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 6.228673328009191e-06,
+      "loss": 1.8629,
+      "step": 3118
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 6.217385313343525e-06,
+      "loss": 1.5028,
+      "step": 3119
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 6.206105653635652e-06,
+      "loss": 1.3017,
+      "step": 3120
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 6.194834355723249e-06,
+      "loss": 1.5684,
+      "step": 3121
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 6.183571426438928e-06,
+      "loss": 1.2778,
+      "step": 3122
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 6.172316872610222e-06,
+      "loss": 1.639,
+      "step": 3123
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 6.16107070105959e-06,
+      "loss": 1.587,
+      "step": 3124
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 6.149832918604422e-06,
+      "loss": 1.4172,
+      "step": 3125
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 6.138603532056988e-06,
+      "loss": 1.6348,
+      "step": 3126
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 6.127382548224503e-06,
+      "loss": 1.3124,
+      "step": 3127
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 6.116169973909077e-06,
+      "loss": 1.3688,
+      "step": 3128
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 6.104965815907715e-06,
+      "loss": 1.5573,
+      "step": 3129
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 6.093770081012334e-06,
+      "loss": 1.8118,
+      "step": 3130
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 6.082582776009729e-06,
+      "loss": 1.8416,
+      "step": 3131
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 6.071403907681599e-06,
+      "loss": 1.544,
+      "step": 3132
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 6.060233482804527e-06,
+      "loss": 1.2007,
+      "step": 3133
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 6.049071508149959e-06,
+      "loss": 1.3516,
+      "step": 3134
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 6.037917990484243e-06,
+      "loss": 1.438,
+      "step": 3135
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 6.0267729365685875e-06,
+      "loss": 1.4501,
+      "step": 3136
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 6.015636353159073e-06,
+      "loss": 1.831,
+      "step": 3137
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 6.004508247006648e-06,
+      "loss": 1.6831,
+      "step": 3138
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 5.993388624857115e-06,
+      "loss": 1.7547,
+      "step": 3139
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.982277493451147e-06,
+      "loss": 1.4519,
+      "step": 3140
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 5.971174859524249e-06,
+      "loss": 1.3019,
+      "step": 3141
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 5.960080729806792e-06,
+      "loss": 1.1684,
+      "step": 3142
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 5.948995111023985e-06,
+      "loss": 1.8403,
+      "step": 3143
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 5.93791800989588e-06,
+      "loss": 1.3309,
+      "step": 3144
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 5.926849433137367e-06,
+      "loss": 1.4393,
+      "step": 3145
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 5.915789387458164e-06,
+      "loss": 1.7554,
+      "step": 3146
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 5.904737879562823e-06,
+      "loss": 1.3628,
+      "step": 3147
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 5.893694916150714e-06,
+      "loss": 1.5982,
+      "step": 3148
+    },
+    {
+      "epoch": 30.06,
+      "learning_rate": 5.882660503916031e-06,
+      "loss": 1.5018,
+      "step": 3149
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 5.871634649547786e-06,
+      "loss": 1.304,
+      "step": 3150
+    },
+    {
+      "epoch": 30.08,
+      "learning_rate": 5.860617359729801e-06,
+      "loss": 1.1965,
+      "step": 3151
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 5.8496086411407135e-06,
+      "loss": 1.6815,
+      "step": 3152
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 5.838608500453941e-06,
+      "loss": 2.0383,
+      "step": 3153
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 5.827616944337733e-06,
+      "loss": 1.3259,
+      "step": 3154
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 5.816633979455111e-06,
+      "loss": 1.2554,
+      "step": 3155
+    },
+    {
+      "epoch": 30.13,
+      "learning_rate": 5.805659612463905e-06,
+      "loss": 1.6812,
+      "step": 3156
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 5.794693850016721e-06,
+      "loss": 1.827,
+      "step": 3157
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 5.783736698760958e-06,
+      "loss": 1.4143,
+      "step": 3158
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 5.77278816533879e-06,
+      "loss": 1.3742,
+      "step": 3159
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 5.761848256387166e-06,
+      "loss": 1.3268,
+      "step": 3160
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 5.750916978537808e-06,
+      "loss": 1.4987,
+      "step": 3161
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 5.7399943384172075e-06,
+      "loss": 1.0731,
+      "step": 3162
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 5.72908034264662e-06,
+      "loss": 1.575,
+      "step": 3163
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 5.718174997842061e-06,
+      "loss": 1.4451,
+      "step": 3164
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 5.707278310614301e-06,
+      "loss": 1.8147,
+      "step": 3165
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 5.696390287568856e-06,
+      "loss": 1.492,
+      "step": 3166
+    },
+    {
+      "epoch": 30.23,
+      "learning_rate": 5.685510935305998e-06,
+      "loss": 1.6562,
+      "step": 3167
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 5.6746402604207405e-06,
+      "loss": 1.537,
+      "step": 3168
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 5.6637782695028374e-06,
+      "loss": 1.394,
+      "step": 3169
+    },
+    {
+      "epoch": 30.26,
+      "learning_rate": 5.65292496913678e-06,
+      "loss": 1.6177,
+      "step": 3170
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 5.642080365901783e-06,
+      "loss": 1.6287,
+      "step": 3171
+    },
+    {
+      "epoch": 30.28,
+      "learning_rate": 5.6312444663718014e-06,
+      "loss": 1.3312,
+      "step": 3172
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 5.620417277115506e-06,
+      "loss": 1.4323,
+      "step": 3173
+    },
+    {
+      "epoch": 30.3,
+      "learning_rate": 5.609598804696288e-06,
+      "loss": 1.4475,
+      "step": 3174
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 5.598789055672258e-06,
+      "loss": 1.2082,
+      "step": 3175
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 5.587988036596232e-06,
+      "loss": 1.4271,
+      "step": 3176
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 5.577195754015743e-06,
+      "loss": 1.4276,
+      "step": 3177
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 5.566412214473021e-06,
+      "loss": 1.4889,
+      "step": 3178
+    },
+    {
+      "epoch": 30.35,
+      "learning_rate": 5.5556374245050045e-06,
+      "loss": 1.7508,
+      "step": 3179
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 5.5448713906433095e-06,
+      "loss": 1.8767,
+      "step": 3180
+    },
+    {
+      "epoch": 30.37,
+      "learning_rate": 5.534114119414262e-06,
+      "loss": 1.5011,
+      "step": 3181
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 5.523365617338872e-06,
+      "loss": 1.9089,
+      "step": 3182
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 5.51262589093283e-06,
+      "loss": 1.6872,
+      "step": 3183
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 5.50189494670651e-06,
+      "loss": 1.5149,
+      "step": 3184
+    },
+    {
+      "epoch": 30.41,
+      "learning_rate": 5.491172791164962e-06,
+      "loss": 1.6405,
+      "step": 3185
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 5.4804594308079075e-06,
+      "loss": 1.5555,
+      "step": 3186
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 5.469754872129734e-06,
+      "loss": 1.5855,
+      "step": 3187
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 5.459059121619496e-06,
+      "loss": 1.2938,
+      "step": 3188
+    },
+    {
+      "epoch": 30.44,
+      "learning_rate": 5.448372185760913e-06,
+      "loss": 1.5827,
+      "step": 3189
+    },
+    {
+      "epoch": 30.45,
+      "learning_rate": 5.437694071032351e-06,
+      "loss": 1.3367,
+      "step": 3190
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 5.427024783906838e-06,
+      "loss": 1.8339,
+      "step": 3191
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 5.416364330852051e-06,
+      "loss": 1.5694,
+      "step": 3192
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 5.405712718330294e-06,
+      "loss": 1.3101,
+      "step": 3193
+    },
+    {
+      "epoch": 30.49,
+      "learning_rate": 5.395069952798535e-06,
+      "loss": 1.505,
+      "step": 3194
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 5.384436040708369e-06,
+      "loss": 1.5151,
+      "step": 3195
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 5.373810988506025e-06,
+      "loss": 1.3818,
+      "step": 3196
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 5.3631948026323585e-06,
+      "loss": 1.9471,
+      "step": 3197
+    },
+    {
+      "epoch": 30.53,
+      "learning_rate": 5.352587489522853e-06,
+      "loss": 1.255,
+      "step": 3198
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 5.341989055607617e-06,
+      "loss": 1.8886,
+      "step": 3199
+    },
+    {
+      "epoch": 30.55,
+      "learning_rate": 5.331399507311368e-06,
+      "loss": 1.5933,
+      "step": 3200
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 5.320818851053444e-06,
+      "loss": 1.9296,
+      "step": 3201
+    },
+    {
+      "epoch": 30.57,
+      "learning_rate": 5.310247093247795e-06,
+      "loss": 1.5586,
+      "step": 3202
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 5.299684240302962e-06,
+      "loss": 1.6083,
+      "step": 3203
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 5.2891302986221e-06,
+      "loss": 1.3259,
+      "step": 3204
+    },
+    {
+      "epoch": 30.6,
+      "learning_rate": 5.278585274602965e-06,
+      "loss": 1.6182,
+      "step": 3205
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 5.2680491746379014e-06,
+      "loss": 1.7023,
+      "step": 3206
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 5.25752200511384e-06,
+      "loss": 1.6414,
+      "step": 3207
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 5.2470037724123065e-06,
+      "loss": 1.5384,
+      "step": 3208
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 5.236494482909411e-06,
+      "loss": 1.5479,
+      "step": 3209
+    },
+    {
+      "epoch": 30.64,
+      "learning_rate": 5.225994142975826e-06,
+      "loss": 1.5854,
+      "step": 3210
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 5.2155027589768114e-06,
+      "loss": 1.5158,
+      "step": 3211
+    },
+    {
+      "epoch": 30.66,
+      "learning_rate": 5.2050203372722e-06,
+      "loss": 1.4292,
+      "step": 3212
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 5.194546884216385e-06,
+      "loss": 1.4009,
+      "step": 3213
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 5.1840824061583265e-06,
+      "loss": 1.3805,
+      "step": 3214
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 5.173626909441542e-06,
+      "loss": 1.6675,
+      "step": 3215
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 5.1631804004041084e-06,
+      "loss": 1.8264,
+      "step": 3216
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 5.15274288537865e-06,
+      "loss": 1.8331,
+      "step": 3217
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 5.14231437069234e-06,
+      "loss": 1.5532,
+      "step": 3218
+    },
+    {
+      "epoch": 30.73,
+      "learning_rate": 5.131894862666902e-06,
+      "loss": 1.8134,
+      "step": 3219
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 5.121484367618585e-06,
+      "loss": 1.5381,
+      "step": 3220
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 5.111082891858186e-06,
+      "loss": 1.9318,
+      "step": 3221
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 5.100690441691032e-06,
+      "loss": 1.4496,
+      "step": 3222
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 5.090307023416981e-06,
+      "loss": 1.5944,
+      "step": 3223
+    },
+    {
+      "epoch": 30.78,
+      "learning_rate": 5.07993264333041e-06,
+      "loss": 1.0359,
+      "step": 3224
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 5.069567307720223e-06,
+      "loss": 1.8135,
+      "step": 3225
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 5.059211022869839e-06,
+      "loss": 1.8525,
+      "step": 3226
+    },
+    {
+      "epoch": 30.81,
+      "learning_rate": 5.048863795057191e-06,
+      "loss": 1.368,
+      "step": 3227
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 5.038525630554718e-06,
+      "loss": 1.7258,
+      "step": 3228
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 5.02819653562937e-06,
+      "loss": 1.2474,
+      "step": 3229
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 5.017876516542599e-06,
+      "loss": 1.6888,
+      "step": 3230
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 5.0075655795503485e-06,
+      "loss": 1.5436,
+      "step": 3231
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 4.997263730903066e-06,
+      "loss": 1.6967,
+      "step": 3232
+    },
+    {
+      "epoch": 30.86,
+      "learning_rate": 4.986970976845687e-06,
+      "loss": 1.5586,
+      "step": 3233
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 4.976687323617622e-06,
+      "loss": 1.4467,
+      "step": 3234
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 4.96641277745278e-06,
+      "loss": 1.8825,
+      "step": 3235
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 4.956147344579543e-06,
+      "loss": 1.6745,
+      "step": 3236
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 4.945891031220771e-06,
+      "loss": 1.9534,
+      "step": 3237
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 4.935643843593792e-06,
+      "loss": 1.4925,
+      "step": 3238
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 4.925405787910405e-06,
+      "loss": 1.7466,
+      "step": 3239
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 4.915176870376872e-06,
+      "loss": 1.3721,
+      "step": 3240
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 4.9049570971939165e-06,
+      "loss": 1.4213,
+      "step": 3241
+    },
+    {
+      "epoch": 30.95,
+      "learning_rate": 4.894746474556717e-06,
+      "loss": 1.7836,
+      "step": 3242
+    },
+    {
+      "epoch": 30.96,
+      "learning_rate": 4.884545008654908e-06,
+      "loss": 1.3595,
+      "step": 3243
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 4.874352705672571e-06,
+      "loss": 1.3384,
+      "step": 3244
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 4.864169571788233e-06,
+      "loss": 1.3423,
+      "step": 3245
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 4.853995613174862e-06,
+      "loss": 1.7847,
+      "step": 3246
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 4.843830835999872e-06,
+      "loss": 1.9301,
+      "step": 3247
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 4.833675246425094e-06,
+      "loss": 1.7643,
+      "step": 3248
+    },
+    {
+      "epoch": 31.02,
+      "learning_rate": 4.8235288506068024e-06,
+      "loss": 1.6659,
+      "step": 3249
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 4.813391654695696e-06,
+      "loss": 1.4888,
+      "step": 3250
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 4.803263664836899e-06,
+      "loss": 1.4817,
+      "step": 3251
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 4.793144887169952e-06,
+      "loss": 1.3309,
+      "step": 3252
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 4.783035327828809e-06,
+      "loss": 1.3576,
+      "step": 3253
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 4.77293499294184e-06,
+      "loss": 1.8649,
+      "step": 3254
+    },
+    {
+      "epoch": 31.07,
+      "learning_rate": 4.7628438886318225e-06,
+      "loss": 1.5513,
+      "step": 3255
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 4.752762021015938e-06,
+      "loss": 1.4117,
+      "step": 3256
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 4.742689396205766e-06,
+      "loss": 1.5338,
+      "step": 3257
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 4.732626020307287e-06,
+      "loss": 1.5556,
+      "step": 3258
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 4.7225718994208756e-06,
+      "loss": 1.4056,
+      "step": 3259
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 4.712527039641295e-06,
+      "loss": 1.7229,
+      "step": 3260
+    },
+    {
+      "epoch": 31.13,
+      "learning_rate": 4.702491447057684e-06,
+      "loss": 1.4433,
+      "step": 3261
+    },
+    {
+      "epoch": 31.14,
+      "learning_rate": 4.692465127753578e-06,
+      "loss": 1.8273,
+      "step": 3262
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 4.682448087806886e-06,
+      "loss": 1.3867,
+      "step": 3263
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 4.672440333289892e-06,
+      "loss": 1.4626,
+      "step": 3264
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 4.66244187026925e-06,
+      "loss": 1.1233,
+      "step": 3265
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 4.652452704805985e-06,
+      "loss": 1.6175,
+      "step": 3266
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 4.642472842955479e-06,
+      "loss": 1.3901,
+      "step": 3267
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 4.632502290767482e-06,
+      "loss": 1.4415,
+      "step": 3268
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 4.622541054286094e-06,
+      "loss": 1.6652,
+      "step": 3269
+    },
+    {
+      "epoch": 31.22,
+      "learning_rate": 4.6125891395497746e-06,
+      "loss": 1.7718,
+      "step": 3270
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 4.602646552591332e-06,
+      "loss": 1.535,
+      "step": 3271
+    },
+    {
+      "epoch": 31.24,
+      "learning_rate": 4.592713299437905e-06,
+      "loss": 1.776,
+      "step": 3272
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 4.582789386110993e-06,
+      "loss": 1.8854,
+      "step": 3273
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 4.572874818626427e-06,
+      "loss": 1.3451,
+      "step": 3274
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 4.562969602994372e-06,
+      "loss": 1.3264,
+      "step": 3275
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 4.553073745219323e-06,
+      "loss": 1.7529,
+      "step": 3276
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 4.543187251300105e-06,
+      "loss": 1.2527,
+      "step": 3277
+    },
+    {
+      "epoch": 31.29,
+      "learning_rate": 4.533310127229871e-06,
+      "loss": 2.1529,
+      "step": 3278
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 4.523442378996077e-06,
+      "loss": 1.265,
+      "step": 3279
+    },
+    {
+      "epoch": 31.31,
+      "learning_rate": 4.51358401258051e-06,
+      "loss": 1.2353,
+      "step": 3280
+    },
+    {
+      "epoch": 31.32,
+      "learning_rate": 4.503735033959269e-06,
+      "loss": 1.2931,
+      "step": 3281
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 4.4938954491027605e-06,
+      "loss": 1.521,
+      "step": 3282
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 4.484065263975692e-06,
+      "loss": 1.9144,
+      "step": 3283
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 4.47424448453708e-06,
+      "loss": 1.7315,
+      "step": 3284
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 4.464433116740237e-06,
+      "loss": 1.4539,
+      "step": 3285
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 4.454631166532764e-06,
+      "loss": 1.6914,
+      "step": 3286
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 4.444838639856568e-06,
+      "loss": 1.4085,
+      "step": 3287
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 4.435055542647822e-06,
+      "loss": 1.5383,
+      "step": 3288
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 4.425281880836998e-06,
+      "loss": 1.9074,
+      "step": 3289
+    },
+    {
+      "epoch": 31.41,
+      "learning_rate": 4.415517660348847e-06,
+      "loss": 1.4394,
+      "step": 3290
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 4.4057628871023935e-06,
+      "loss": 1.3429,
+      "step": 3291
+    },
+    {
+      "epoch": 31.43,
+      "learning_rate": 4.396017567010939e-06,
+      "loss": 1.8038,
+      "step": 3292
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 4.38628170598205e-06,
+      "loss": 1.8812,
+      "step": 3293
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 4.376555309917559e-06,
+      "loss": 1.1976,
+      "step": 3294
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 4.366838384713566e-06,
+      "loss": 1.2403,
+      "step": 3295
+    },
+    {
+      "epoch": 31.47,
+      "learning_rate": 4.357130936260423e-06,
+      "loss": 1.7486,
+      "step": 3296
+    },
+    {
+      "epoch": 31.47,
+      "learning_rate": 4.347432970442742e-06,
+      "loss": 1.7679,
+      "step": 3297
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 4.337744493139386e-06,
+      "loss": 1.7652,
+      "step": 3298
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 4.328065510223467e-06,
+      "loss": 1.7415,
+      "step": 3299
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 4.318396027562337e-06,
+      "loss": 1.6109,
+      "step": 3300
+    },
+    {
+      "epoch": 31.51,
+      "learning_rate": 4.308736051017601e-06,
+      "loss": 1.2971,
+      "step": 3301
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 4.299085586445078e-06,
+      "loss": 1.5768,
+      "step": 3302
+    },
+    {
+      "epoch": 31.53,
+      "learning_rate": 4.289444639694842e-06,
+      "loss": 1.5397,
+      "step": 3303
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 4.2798132166111925e-06,
+      "loss": 1.8041,
+      "step": 3304
+    },
+    {
+      "epoch": 31.55,
+      "learning_rate": 4.270191323032651e-06,
+      "loss": 1.6378,
+      "step": 3305
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 4.260578964791966e-06,
+      "loss": 1.6802,
+      "step": 3306
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 4.250976147716108e-06,
+      "loss": 1.66,
+      "step": 3307
+    },
+    {
+      "epoch": 31.58,
+      "learning_rate": 4.2413828776262566e-06,
+      "loss": 1.3783,
+      "step": 3308
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 4.2317991603378105e-06,
+      "loss": 1.4634,
+      "step": 3309
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 4.2222250016603735e-06,
+      "loss": 1.4915,
+      "step": 3310
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 4.2126604073977594e-06,
+      "loss": 1.548,
+      "step": 3311
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 4.203105383347978e-06,
+      "loss": 1.2589,
+      "step": 3312
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 4.193559935303244e-06,
+      "loss": 1.428,
+      "step": 3313
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 4.184024069049968e-06,
+      "loss": 1.6164,
+      "step": 3314
+    },
+    {
+      "epoch": 31.65,
+      "learning_rate": 4.1744977903687366e-06,
+      "loss": 1.8102,
+      "step": 3315
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 4.164981105034342e-06,
+      "loss": 1.6374,
+      "step": 3316
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 4.1554740188157595e-06,
+      "loss": 1.7421,
+      "step": 3317
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 4.1459765374761355e-06,
+      "loss": 1.4112,
+      "step": 3318
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 4.136488666772802e-06,
+      "loss": 1.682,
+      "step": 3319
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 4.1270104124572645e-06,
+      "loss": 1.9241,
+      "step": 3320
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 4.117541780275191e-06,
+      "loss": 1.654,
+      "step": 3321
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 4.108082775966431e-06,
+      "loss": 1.8491,
+      "step": 3322
+    },
+    {
+      "epoch": 31.72,
+      "learning_rate": 4.098633405264985e-06,
+      "loss": 1.8634,
+      "step": 3323
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 4.089193673899019e-06,
+      "loss": 1.3985,
+      "step": 3324
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 4.079763587590855e-06,
+      "loss": 1.5794,
+      "step": 3325
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 4.070343152056966e-06,
+      "loss": 1.1574,
+      "step": 3326
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 4.06093237300798e-06,
+      "loss": 1.4454,
+      "step": 3327
+    },
+    {
+      "epoch": 31.77,
+      "learning_rate": 4.051531256148671e-06,
+      "loss": 1.5718,
+      "step": 3328
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 4.0421398071779446e-06,
+      "loss": 1.5705,
+      "step": 3329
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 4.032758031788857e-06,
+      "loss": 1.523,
+      "step": 3330
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 4.0233859356686e-06,
+      "loss": 1.4256,
+      "step": 3331
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 4.014023524498492e-06,
+      "loss": 1.4341,
+      "step": 3332
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 4.004670803953987e-06,
+      "loss": 1.4973,
+      "step": 3333
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 3.995327779704658e-06,
+      "loss": 1.4428,
+      "step": 3334
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 3.985994457414206e-06,
+      "loss": 2.0886,
+      "step": 3335
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 3.9766708427404485e-06,
+      "loss": 1.4529,
+      "step": 3336
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 3.9673569413353165e-06,
+      "loss": 1.3978,
+      "step": 3337
+    },
+    {
+      "epoch": 31.87,
+      "learning_rate": 3.958052758844853e-06,
+      "loss": 1.4526,
+      "step": 3338
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 3.948758300909212e-06,
+      "loss": 1.6905,
+      "step": 3339
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 3.93947357316266e-06,
+      "loss": 1.4096,
+      "step": 3340
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 3.93019858123354e-06,
+      "loss": 1.774,
+      "step": 3341
+    },
+    {
+      "epoch": 31.9,
+      "learning_rate": 3.920933330744319e-06,
+      "loss": 1.7426,
+      "step": 3342
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 3.911677827311547e-06,
+      "loss": 1.3783,
+      "step": 3343
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 3.902432076545868e-06,
+      "loss": 1.5029,
+      "step": 3344
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 3.893196084052013e-06,
+      "loss": 1.4134,
+      "step": 3345
+    },
+    {
+      "epoch": 31.94,
+      "learning_rate": 3.883969855428802e-06,
+      "loss": 1.8311,
+      "step": 3346
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 3.874753396269135e-06,
+      "loss": 1.2843,
+      "step": 3347
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 3.865546712159978e-06,
+      "loss": 1.3759,
+      "step": 3348
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 3.8563498086823825e-06,
+      "loss": 1.8211,
+      "step": 3349
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 3.847162691411472e-06,
+      "loss": 1.3907,
+      "step": 3350
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 3.837985365916437e-06,
+      "loss": 1.4587,
+      "step": 3351
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 3.828817837760525e-06,
+      "loss": 1.7053,
+      "step": 3352
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 3.819660112501053e-06,
+      "loss": 1.4964,
+      "step": 3353
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 3.810512195689391e-06,
+      "loss": 1.3375,
+      "step": 3354
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 3.8013740928709707e-06,
+      "loss": 1.9915,
+      "step": 3355
+    },
+    {
+      "epoch": 32.04,
+      "learning_rate": 3.7922458095852556e-06,
+      "loss": 2.0446,
+      "step": 3356
+    },
+    {
+      "epoch": 32.05,
+      "learning_rate": 3.783127351365774e-06,
+      "loss": 1.5737,
+      "step": 3357
+    },
+    {
+      "epoch": 32.06,
+      "learning_rate": 3.774018723740096e-06,
+      "loss": 1.4806,
+      "step": 3358
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 3.7649199322298293e-06,
+      "loss": 1.6114,
+      "step": 3359
+    },
+    {
+      "epoch": 32.08,
+      "learning_rate": 3.7558309823506166e-06,
+      "loss": 1.2261,
+      "step": 3360
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 3.7467518796121426e-06,
+      "loss": 1.5252,
+      "step": 3361
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 3.7376826295181133e-06,
+      "loss": 1.6914,
+      "step": 3362
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 3.728623237566269e-06,
+      "loss": 1.4126,
+      "step": 3363
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 3.7195737092483697e-06,
+      "loss": 1.4021,
+      "step": 3364
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 3.7105340500502006e-06,
+      "loss": 1.7246,
+      "step": 3365
+    },
+    {
+      "epoch": 32.13,
+      "learning_rate": 3.7015042654515585e-06,
+      "loss": 1.045,
+      "step": 3366
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 3.6924843609262604e-06,
+      "loss": 1.6685,
+      "step": 3367
+    },
+    {
+      "epoch": 32.15,
+      "learning_rate": 3.683474341942135e-06,
+      "loss": 1.2006,
+      "step": 3368
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 3.6744742139610034e-06,
+      "loss": 1.3087,
+      "step": 3369
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 3.6654839824387068e-06,
+      "loss": 1.3929,
+      "step": 3370
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 3.656503652825083e-06,
+      "loss": 1.7689,
+      "step": 3371
+    },
+    {
+      "epoch": 32.19,
+      "learning_rate": 3.647533230563969e-06,
+      "loss": 1.3003,
+      "step": 3372
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 3.63857272109319e-06,
+      "loss": 1.6693,
+      "step": 3373
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 3.6296221298445656e-06,
+      "loss": 1.593,
+      "step": 3374
+    },
+    {
+      "epoch": 32.22,
+      "learning_rate": 3.6206814622439046e-06,
+      "loss": 1.5685,
+      "step": 3375
+    },
+    {
+      "epoch": 32.23,
+      "learning_rate": 3.6117507237110007e-06,
+      "loss": 1.4985,
+      "step": 3376
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 3.602829919659623e-06,
+      "loss": 1.1429,
+      "step": 3377
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 3.5939190554975258e-06,
+      "loss": 1.2295,
+      "step": 3378
+    },
+    {
+      "epoch": 32.26,
+      "learning_rate": 3.58501813662643e-06,
+      "loss": 1.4901,
+      "step": 3379
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 3.576127168442034e-06,
+      "loss": 1.9365,
+      "step": 3380
+    },
+    {
+      "epoch": 32.28,
+      "learning_rate": 3.5672461563340012e-06,
+      "loss": 1.7289,
+      "step": 3381
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 3.558375105685965e-06,
+      "loss": 1.6749,
+      "step": 3382
+    },
+    {
+      "epoch": 32.3,
+      "learning_rate": 3.5495140218755085e-06,
+      "loss": 1.9555,
+      "step": 3383
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 3.5406629102741818e-06,
+      "loss": 1.3574,
+      "step": 3384
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 3.5318217762474903e-06,
+      "loss": 1.0681,
+      "step": 3385
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 3.5229906251548873e-06,
+      "loss": 1.6297,
+      "step": 3386
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 3.514169462349777e-06,
+      "loss": 1.7338,
+      "step": 3387
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 3.505358293179506e-06,
+      "loss": 2.0013,
+      "step": 3388
+    },
+    {
+      "epoch": 32.35,
+      "learning_rate": 3.4965571229853668e-06,
+      "loss": 1.3415,
+      "step": 3389
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 3.4877659571025866e-06,
+      "loss": 1.5564,
+      "step": 3390
+    },
+    {
+      "epoch": 32.37,
+      "learning_rate": 3.4789848008603343e-06,
+      "loss": 1.7575,
+      "step": 3391
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 3.4702136595817002e-06,
+      "loss": 1.4984,
+      "step": 3392
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 3.461452538583714e-06,
+      "loss": 1.3821,
+      "step": 3393
+    },
+    {
+      "epoch": 32.4,
+      "learning_rate": 3.4527014431773266e-06,
+      "loss": 1.3558,
+      "step": 3394
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 3.4439603786674146e-06,
+      "loss": 1.6439,
+      "step": 3395
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 3.4352293503527622e-06,
+      "loss": 1.7785,
+      "step": 3396
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 3.4265083635260863e-06,
+      "loss": 1.5199,
+      "step": 3397
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 3.4177974234740075e-06,
+      "loss": 1.7578,
+      "step": 3398
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 3.4090965354770565e-06,
+      "loss": 1.6274,
+      "step": 3399
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 3.4004057048096708e-06,
+      "loss": 1.5922,
+      "step": 3400
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 3.391724936740195e-06,
+      "loss": 1.6803,
+      "step": 3401
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 3.383054236530872e-06,
+      "loss": 1.7133,
+      "step": 3402
+    },
+    {
+      "epoch": 32.49,
+      "learning_rate": 3.374393609437836e-06,
+      "loss": 1.5997,
+      "step": 3403
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 3.365743060711122e-06,
+      "loss": 1.5228,
+      "step": 3404
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 3.35710259559465e-06,
+      "loss": 1.5918,
+      "step": 3405
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 3.3484722193262355e-06,
+      "loss": 1.476,
+      "step": 3406
+    },
+    {
+      "epoch": 32.53,
+      "learning_rate": 3.3398519371375705e-06,
+      "loss": 1.4023,
+      "step": 3407
+    },
+    {
+      "epoch": 32.53,
+      "learning_rate": 3.3312417542542306e-06,
+      "loss": 1.6594,
+      "step": 3408
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 3.3226416758956724e-06,
+      "loss": 1.8098,
+      "step": 3409
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 3.3140517072752167e-06,
+      "loss": 1.7661,
+      "step": 3410
+    },
+    {
+      "epoch": 32.56,
+      "learning_rate": 3.3054718536000663e-06,
+      "loss": 1.5452,
+      "step": 3411
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 3.296902120071288e-06,
+      "loss": 1.5892,
+      "step": 3412
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 3.2883425118838197e-06,
+      "loss": 1.2072,
+      "step": 3413
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 3.279793034226453e-06,
+      "loss": 1.9333,
+      "step": 3414
+    },
+    {
+      "epoch": 32.6,
+      "learning_rate": 3.271253692281842e-06,
+      "loss": 1.5336,
+      "step": 3415
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 3.262724491226501e-06,
+      "loss": 1.0076,
+      "step": 3416
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 3.254205436230784e-06,
+      "loss": 1.6171,
+      "step": 3417
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 3.2456965324589083e-06,
+      "loss": 1.7611,
+      "step": 3418
+    },
+    {
+      "epoch": 32.64,
+      "learning_rate": 3.237197785068933e-06,
+      "loss": 1.1915,
+      "step": 3419
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 3.228709199212756e-06,
+      "loss": 1.3708,
+      "step": 3420
+    },
+    {
+      "epoch": 32.66,
+      "learning_rate": 3.220230780036124e-06,
+      "loss": 1.518,
+      "step": 3421
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 3.2117625326786127e-06,
+      "loss": 1.7983,
+      "step": 3422
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 3.2033044622736355e-06,
+      "loss": 1.985,
+      "step": 3423
+    },
+    {
+      "epoch": 32.69,
+      "learning_rate": 3.1948565739484414e-06,
+      "loss": 1.7903,
+      "step": 3424
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 3.1864188728240906e-06,
+      "loss": 1.3773,
+      "step": 3425
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 3.1779913640154845e-06,
+      "loss": 1.3231,
+      "step": 3426
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 3.169574052631341e-06,
+      "loss": 1.6048,
+      "step": 3427
+    },
+    {
+      "epoch": 32.73,
+      "learning_rate": 3.1611669437741944e-06,
+      "loss": 1.8171,
+      "step": 3428
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 3.152770042540394e-06,
+      "loss": 1.5551,
+      "step": 3429
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 3.1443833540201063e-06,
+      "loss": 1.5345,
+      "step": 3430
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 3.1360068832973024e-06,
+      "loss": 1.6968,
+      "step": 3431
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 3.127640635449758e-06,
+      "loss": 1.7957,
+      "step": 3432
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 3.1192846155490587e-06,
+      "loss": 1.3418,
+      "step": 3433
+    },
+    {
+      "epoch": 32.78,
+      "learning_rate": 3.1109388286605815e-06,
+      "loss": 1.5259,
+      "step": 3434
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 3.1026032798435037e-06,
+      "loss": 1.8731,
+      "step": 3435
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 3.0942779741508054e-06,
+      "loss": 1.5523,
+      "step": 3436
+    },
+    {
+      "epoch": 32.81,
+      "learning_rate": 3.085962916629235e-06,
+      "loss": 1.371,
+      "step": 3437
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 3.0776581123193503e-06,
+      "loss": 1.2995,
+      "step": 3438
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 3.0693635662554833e-06,
+      "loss": 1.7661,
+      "step": 3439
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 3.061079283465751e-06,
+      "loss": 1.3086,
+      "step": 3440
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 3.0528052689720477e-06,
+      "loss": 1.3748,
+      "step": 3441
+    },
+    {
+      "epoch": 32.86,
+      "learning_rate": 3.0445415277900435e-06,
+      "loss": 1.5753,
+      "step": 3442
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 3.036288064929178e-06,
+      "loss": 1.4242,
+      "step": 3443
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 3.0280448853926647e-06,
+      "loss": 1.468,
+      "step": 3444
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 3.0198119941774796e-06,
+      "loss": 1.7216,
+      "step": 3445
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 3.011589396274366e-06,
+      "loss": 1.2284,
+      "step": 3446
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 3.0033770966678235e-06,
+      "loss": 1.9973,
+      "step": 3447
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 2.9951751003361095e-06,
+      "loss": 1.3587,
+      "step": 3448
+    },
+    {
+      "epoch": 32.93,
+      "learning_rate": 2.9869834122512388e-06,
+      "loss": 1.8085,
+      "step": 3449
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 2.978802037378976e-06,
+      "loss": 1.4045,
+      "step": 3450
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 2.970630980678828e-06,
+      "loss": 1.6745,
+      "step": 3451
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 2.962470247104057e-06,
+      "loss": 1.489,
+      "step": 3452
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 2.954319841601656e-06,
+      "loss": 1.2512,
+      "step": 3453
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 2.9461797691123696e-06,
+      "loss": 1.8907,
+      "step": 3454
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 2.93805003457067e-06,
+      "loss": 1.3068,
+      "step": 3455
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 2.929930642904766e-06,
+      "loss": 1.4232,
+      "step": 3456
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 2.9218215990365938e-06,
+      "loss": 1.6741,
+      "step": 3457
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 2.9137229078818197e-06,
+      "loss": 1.7144,
+      "step": 3458
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 2.905634574349834e-06,
+      "loss": 1.3452,
+      "step": 3459
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 2.8975566033437453e-06,
+      "loss": 1.907,
+      "step": 3460
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 2.8894889997603858e-06,
+      "loss": 1.5504,
+      "step": 3461
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 2.881431768490295e-06,
+      "loss": 1.4516,
+      "step": 3462
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 2.8733849144177384e-06,
+      "loss": 1.3551,
+      "step": 3463
+    },
+    {
+      "epoch": 33.07,
+      "learning_rate": 2.86534844242067e-06,
+      "loss": 1.7436,
+      "step": 3464
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 2.8573223573707686e-06,
+      "loss": 1.7131,
+      "step": 3465
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 2.849306664133411e-06,
+      "loss": 2.1389,
+      "step": 3466
+    },
+    {
+      "epoch": 33.1,
+      "learning_rate": 2.8413013675676703e-06,
+      "loss": 1.3566,
+      "step": 3467
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 2.833306472526325e-06,
+      "loss": 1.6804,
+      "step": 3468
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 2.82532198385584e-06,
+      "loss": 1.7234,
+      "step": 3469
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 2.817347906396377e-06,
+      "loss": 1.77,
+      "step": 3470
+    },
+    {
+      "epoch": 33.14,
+      "learning_rate": 2.8093842449817855e-06,
+      "loss": 1.8393,
+      "step": 3471
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 2.801431004439599e-06,
+      "loss": 1.7286,
+      "step": 3472
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 2.7934881895910383e-06,
+      "loss": 1.814,
+      "step": 3473
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 2.785555805250999e-06,
+      "loss": 1.5789,
+      "step": 3474
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 2.7776338562280567e-06,
+      "loss": 1.5051,
+      "step": 3475
+    },
+    {
+      "epoch": 33.18,
+      "learning_rate": 2.7697223473244596e-06,
+      "loss": 1.4728,
+      "step": 3476
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 2.7618212833361346e-06,
+      "loss": 1.4839,
+      "step": 3477
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 2.7539306690526602e-06,
+      "loss": 1.9264,
+      "step": 3478
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 2.7460505092572966e-06,
+      "loss": 1.5691,
+      "step": 3479
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 2.7381808087269577e-06,
+      "loss": 1.8107,
+      "step": 3480
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 2.730321572232224e-06,
+      "loss": 1.9962,
+      "step": 3481
+    },
+    {
+      "epoch": 33.24,
+      "learning_rate": 2.722472804537324e-06,
+      "loss": 1.6718,
+      "step": 3482
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 2.714634510400147e-06,
+      "loss": 1.3443,
+      "step": 3483
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 2.7068066945722305e-06,
+      "loss": 1.6399,
+      "step": 3484
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 2.6989893617987607e-06,
+      "loss": 1.5378,
+      "step": 3485
+    },
+    {
+      "epoch": 33.28,
+      "learning_rate": 2.6911825168185733e-06,
+      "loss": 1.2657,
+      "step": 3486
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 2.6833861643641325e-06,
+      "loss": 1.67,
+      "step": 3487
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 2.675600309161559e-06,
+      "loss": 1.3467,
+      "step": 3488
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 2.667824955930598e-06,
+      "loss": 1.6585,
+      "step": 3489
+    },
+    {
+      "epoch": 33.32,
+      "learning_rate": 2.6600601093846345e-06,
+      "loss": 1.5142,
+      "step": 3490
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 2.6523057742306835e-06,
+      "loss": 1.141,
+      "step": 3491
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 2.644561955169387e-06,
+      "loss": 1.7084,
+      "step": 3492
+    },
+    {
+      "epoch": 33.35,
+      "learning_rate": 2.636828656895016e-06,
+      "loss": 1.5579,
+      "step": 3493
+    },
+    {
+      "epoch": 33.36,
+      "learning_rate": 2.6291058840954552e-06,
+      "loss": 1.0198,
+      "step": 3494
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 2.6213936414522145e-06,
+      "loss": 1.5797,
+      "step": 3495
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 2.6136919336404208e-06,
+      "loss": 1.7522,
+      "step": 3496
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 2.6060007653288155e-06,
+      "loss": 1.5417,
+      "step": 3497
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 2.5983201411797487e-06,
+      "loss": 1.5637,
+      "step": 3498
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 2.5906500658491807e-06,
+      "loss": 1.5021,
+      "step": 3499
+    },
+    {
+      "epoch": 33.41,
+      "learning_rate": 2.582990543986674e-06,
+      "loss": 1.4073,
+      "step": 3500
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 2.5753415802353953e-06,
+      "loss": 1.8385,
+      "step": 3501
+    },
+    {
+      "epoch": 33.43,
+      "learning_rate": 2.567703179232115e-06,
+      "loss": 1.5625,
+      "step": 3502
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 2.5600753456071914e-06,
+      "loss": 1.4872,
+      "step": 3503
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 2.5524580839845926e-06,
+      "loss": 1.994,
+      "step": 3504
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 2.5448513989818536e-06,
+      "loss": 1.6635,
+      "step": 3505
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 2.537255295210119e-06,
+      "loss": 1.5723,
+      "step": 3506
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 2.529669777274111e-06,
+      "loss": 1.335,
+      "step": 3507
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 2.5220948497721364e-06,
+      "loss": 1.239,
+      "step": 3508
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 2.514530517296083e-06,
+      "loss": 1.7754,
+      "step": 3509
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 2.506976784431412e-06,
+      "loss": 1.103,
+      "step": 3510
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 2.499433655757162e-06,
+      "loss": 1.5424,
+      "step": 3511
+    },
+    {
+      "epoch": 33.53,
+      "learning_rate": 2.4919011358459443e-06,
+      "loss": 1.7329,
+      "step": 3512
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 2.4843792292639357e-06,
+      "loss": 2.0781,
+      "step": 3513
+    },
+    {
+      "epoch": 33.55,
+      "learning_rate": 2.476867940570884e-06,
+      "loss": 1.2891,
+      "step": 3514
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 2.469367274320096e-06,
+      "loss": 1.768,
+      "step": 3515
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 2.461877235058443e-06,
+      "loss": 1.6541,
+      "step": 3516
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 2.454397827326356e-06,
+      "loss": 1.7575,
+      "step": 3517
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 2.4469290556578073e-06,
+      "loss": 1.1775,
+      "step": 3518
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 2.4394709245803384e-06,
+      "loss": 1.909,
+      "step": 3519
+    },
+    {
+      "epoch": 33.6,
+      "learning_rate": 2.4320234386150343e-06,
+      "loss": 1.6309,
+      "step": 3520
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 2.424586602276526e-06,
+      "loss": 1.2455,
+      "step": 3521
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 2.4171604200729883e-06,
+      "loss": 1.7379,
+      "step": 3522
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 2.409744896506141e-06,
+      "loss": 1.4047,
+      "step": 3523
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 2.40234003607124e-06,
+      "loss": 1.5599,
+      "step": 3524
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 2.394945843257077e-06,
+      "loss": 1.683,
+      "step": 3525
+    },
+    {
+      "epoch": 33.66,
+      "learning_rate": 2.3875623225459777e-06,
+      "loss": 1.5814,
+      "step": 3526
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 2.380189478413799e-06,
+      "loss": 1.0929,
+      "step": 3527
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 2.372827315329924e-06,
+      "loss": 1.8063,
+      "step": 3528
+    },
+    {
+      "epoch": 33.69,
+      "learning_rate": 2.365475837757265e-06,
+      "loss": 1.4389,
+      "step": 3529
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 2.3581350501522527e-06,
+      "loss": 1.4081,
+      "step": 3530
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 2.3508049569648406e-06,
+      "loss": 1.5232,
+      "step": 3531
+    },
+    {
+      "epoch": 33.72,
+      "learning_rate": 2.343485562638492e-06,
+      "loss": 1.496,
+      "step": 3532
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 2.336176871610194e-06,
+      "loss": 1.865,
+      "step": 3533
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 2.3288788883104396e-06,
+      "loss": 1.482,
+      "step": 3534
+    },
+    {
+      "epoch": 33.75,
+      "learning_rate": 2.3215916171632346e-06,
+      "loss": 1.2972,
+      "step": 3535
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 2.31431506258609e-06,
+      "loss": 1.474,
+      "step": 3536
+    },
+    {
+      "epoch": 33.77,
+      "learning_rate": 2.3070492289900193e-06,
+      "loss": 1.4877,
+      "step": 3537
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 2.299794120779535e-06,
+      "loss": 1.5263,
+      "step": 3538
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 2.2925497423526544e-06,
+      "loss": 1.4917,
+      "step": 3539
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 2.2853160981008826e-06,
+      "loss": 1.9012,
+      "step": 3540
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 2.2780931924092255e-06,
+      "loss": 1.5346,
+      "step": 3541
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 2.2708810296561713e-06,
+      "loss": 1.6766,
+      "step": 3542
+    },
+    {
+      "epoch": 33.82,
+      "learning_rate": 2.263679614213703e-06,
+      "loss": 1.2229,
+      "step": 3543
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 2.2564889504472887e-06,
+      "loss": 1.3508,
+      "step": 3544
+    },
+    {
+      "epoch": 33.84,
+      "learning_rate": 2.249309042715868e-06,
+      "loss": 1.8268,
+      "step": 3545
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 2.242139895371871e-06,
+      "loss": 1.9762,
+      "step": 3546
+    },
+    {
+      "epoch": 33.86,
+      "learning_rate": 2.2349815127612027e-06,
+      "loss": 1.3359,
+      "step": 3547
+    },
+    {
+      "epoch": 33.87,
+      "learning_rate": 2.227833899223244e-06,
+      "loss": 1.1742,
+      "step": 3548
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 2.2206970590908394e-06,
+      "loss": 1.4588,
+      "step": 3549
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 2.213570996690315e-06,
+      "loss": 1.6209,
+      "step": 3550
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 2.2064557163414536e-06,
+      "loss": 1.5856,
+      "step": 3551
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 2.1993512223575063e-06,
+      "loss": 1.5606,
+      "step": 3552
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 2.1922575190451866e-06,
+      "loss": 1.3436,
+      "step": 3553
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 2.1851746107046614e-06,
+      "loss": 1.6233,
+      "step": 3554
+    },
+    {
+      "epoch": 33.94,
+      "learning_rate": 2.1781025016295645e-06,
+      "loss": 1.1532,
+      "step": 3555
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 2.1710411961069667e-06,
+      "loss": 1.4576,
+      "step": 3556
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 2.163990698417402e-06,
+      "loss": 1.5485,
+      "step": 3557
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 2.1569510128348492e-06,
+      "loss": 1.447,
+      "step": 3558
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 2.149922143626737e-06,
+      "loss": 1.2308,
+      "step": 3559
+    },
+    {
+      "epoch": 33.99,
+      "learning_rate": 2.1429040950539306e-06,
+      "loss": 1.3208,
+      "step": 3560
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 2.1358968713707395e-06,
+      "loss": 1.587,
+      "step": 3561
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 2.1289004768249153e-06,
+      "loss": 1.6269,
+      "step": 3562
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 2.1219149156576325e-06,
+      "loss": 2.0766,
+      "step": 3563
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 2.1149401921035095e-06,
+      "loss": 1.237,
+      "step": 3564
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 2.107976310390589e-06,
+      "loss": 1.723,
+      "step": 3565
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 2.1010232747403502e-06,
+      "loss": 1.4821,
+      "step": 3566
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 2.0940810893676855e-06,
+      "loss": 1.7969,
+      "step": 3567
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 2.0871497584809197e-06,
+      "loss": 1.5575,
+      "step": 3568
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 2.080229286281794e-06,
+      "loss": 1.5945,
+      "step": 3569
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 2.073319676965464e-06,
+      "loss": 1.4269,
+      "step": 3570
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 2.0664209347205076e-06,
+      "loss": 1.4933,
+      "step": 3571
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 2.0595330637289046e-06,
+      "loss": 1.7216,
+      "step": 3572
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 2.0526560681660546e-06,
+      "loss": 1.307,
+      "step": 3573
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 2.0457899522007586e-06,
+      "loss": 1.5622,
+      "step": 3574
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 2.0389347199952268e-06,
+      "loss": 1.6227,
+      "step": 3575
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 2.0320903757050693e-06,
+      "loss": 1.5585,
+      "step": 3576
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 2.0252569234792928e-06,
+      "loss": 1.7718,
+      "step": 3577
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 2.0184343674603114e-06,
+      "loss": 1.6981,
+      "step": 3578
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 2.0116227117839204e-06,
+      "loss": 1.4992,
+      "step": 3579
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 2.0048219605793196e-06,
+      "loss": 1.666,
+      "step": 3580
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 1.998032117969089e-06,
+      "loss": 1.5526,
+      "step": 3581
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 1.991253188069202e-06,
+      "loss": 1.2979,
+      "step": 3582
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 1.984485174989017e-06,
+      "loss": 1.5614,
+      "step": 3583
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 1.9777280828312673e-06,
+      "loss": 1.7111,
+      "step": 3584
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 1.9709819156920784e-06,
+      "loss": 1.4925,
+      "step": 3585
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 1.9642466776609394e-06,
+      "loss": 1.5651,
+      "step": 3586
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 1.9575223728207217e-06,
+      "loss": 1.572,
+      "step": 3587
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 1.950809005247669e-06,
+      "loss": 1.3976,
+      "step": 3588
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 1.9441065790113934e-06,
+      "loss": 1.5912,
+      "step": 3589
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 1.9374150981748752e-06,
+      "loss": 1.7594,
+      "step": 3590
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 1.9307345667944612e-06,
+      "loss": 1.6599,
+      "step": 3591
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 1.9240649889198557e-06,
+      "loss": 1.4027,
+      "step": 3592
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 1.9174063685941304e-06,
+      "loss": 1.5343,
+      "step": 3593
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 1.9107587098537063e-06,
+      "loss": 1.3408,
+      "step": 3594
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 1.9041220167283646e-06,
+      "loss": 1.2679,
+      "step": 3595
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 1.8974962932412412e-06,
+      "loss": 1.843,
+      "step": 3596
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 1.8908815434088156e-06,
+      "loss": 1.5334,
+      "step": 3597
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 1.884277771240921e-06,
+      "loss": 1.1995,
+      "step": 3598
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 1.877684980740735e-06,
+      "loss": 1.4137,
+      "step": 3599
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 1.8711031759047715e-06,
+      "loss": 1.4169,
+      "step": 3600
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 1.864532360722895e-06,
+      "loss": 1.8764,
+      "step": 3601
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 1.857972539178301e-06,
+      "loss": 1.6034,
+      "step": 3602
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 1.8514237152475267e-06,
+      "loss": 1.0766,
+      "step": 3603
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 1.8448858929004343e-06,
+      "loss": 1.6912,
+      "step": 3604
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 1.8383590761002267e-06,
+      "loss": 1.6363,
+      "step": 3605
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 1.831843268803426e-06,
+      "loss": 1.4872,
+      "step": 3606
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 1.825338474959888e-06,
+      "loss": 1.7709,
+      "step": 3607
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 1.8188446985127872e-06,
+      "loss": 1.5696,
+      "step": 3608
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 1.8123619433986217e-06,
+      "loss": 1.4875,
+      "step": 3609
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 1.8058902135472078e-06,
+      "loss": 1.6753,
+      "step": 3610
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 1.7994295128816785e-06,
+      "loss": 1.4559,
+      "step": 3611
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 1.7929798453184832e-06,
+      "loss": 1.3031,
+      "step": 3612
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 1.7865412147673767e-06,
+      "loss": 1.431,
+      "step": 3613
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 1.7801136251314276e-06,
+      "loss": 1.5248,
+      "step": 3614
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 1.7736970803070107e-06,
+      "loss": 1.2447,
+      "step": 3615
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 1.7672915841838101e-06,
+      "loss": 1.5937,
+      "step": 3616
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 1.7608971406448061e-06,
+      "loss": 1.1901,
+      "step": 3617
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 1.7545137535662804e-06,
+      "loss": 1.5031,
+      "step": 3618
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 1.7481414268178132e-06,
+      "loss": 1.7662,
+      "step": 3619
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 1.741780164262281e-06,
+      "loss": 1.5752,
+      "step": 3620
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 1.7354299697558508e-06,
+      "loss": 1.8289,
+      "step": 3621
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 1.7290908471479805e-06,
+      "loss": 1.5647,
+      "step": 3622
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 1.7227628002814189e-06,
+      "loss": 1.4851,
+      "step": 3623
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 1.7164458329922017e-06,
+      "loss": 1.4603,
+      "step": 3624
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 1.7101399491096392e-06,
+      "loss": 1.4801,
+      "step": 3625
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 1.7038451524563316e-06,
+      "loss": 1.2433,
+      "step": 3626
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 1.697561446848155e-06,
+      "loss": 1.6114,
+      "step": 3627
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 1.6912888360942669e-06,
+      "loss": 1.5448,
+      "step": 3628
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 1.6850273239970904e-06,
+      "loss": 1.7876,
+      "step": 3629
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 1.678776914352329e-06,
+      "loss": 1.6564,
+      "step": 3630
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 1.6725376109489521e-06,
+      "loss": 1.6167,
+      "step": 3631
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 1.666309417569194e-06,
+      "loss": 1.6816,
+      "step": 3632
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 1.660092337988557e-06,
+      "loss": 1.7316,
+      "step": 3633
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 1.6538863759758105e-06,
+      "loss": 1.3926,
+      "step": 3634
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 1.6476915352929745e-06,
+      "loss": 1.2744,
+      "step": 3635
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 1.6415078196953382e-06,
+      "loss": 1.8017,
+      "step": 3636
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 1.6353352329314365e-06,
+      "loss": 1.7104,
+      "step": 3637
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 1.6291737787430672e-06,
+      "loss": 1.4245,
+      "step": 3638
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 1.6230234608652784e-06,
+      "loss": 1.8392,
+      "step": 3639
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 1.6168842830263586e-06,
+      "loss": 1.6812,
+      "step": 3640
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 1.610756248947849e-06,
+      "loss": 1.6428,
+      "step": 3641
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 1.6046393623445378e-06,
+      "loss": 1.3948,
+      "step": 3642
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 1.5985336269244545e-06,
+      "loss": 1.472,
+      "step": 3643
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 1.5924390463888673e-06,
+      "loss": 1.7621,
+      "step": 3644
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 1.586355624432283e-06,
+      "loss": 1.484,
+      "step": 3645
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 1.5802833647424453e-06,
+      "loss": 1.8402,
+      "step": 3646
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 1.5742222710003252e-06,
+      "loss": 1.6273,
+      "step": 3647
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 1.5681723468801368e-06,
+      "loss": 2.245,
+      "step": 3648
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 1.5621335960493112e-06,
+      "loss": 1.4299,
+      "step": 3649
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 1.5561060221685132e-06,
+      "loss": 1.0581,
+      "step": 3650
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 1.5500896288916312e-06,
+      "loss": 1.8878,
+      "step": 3651
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 1.5440844198657745e-06,
+      "loss": 1.4895,
+      "step": 3652
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 1.5380903987312756e-06,
+      "loss": 1.313,
+      "step": 3653
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 1.5321075691216792e-06,
+      "loss": 1.243,
+      "step": 3654
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 1.5261359346637484e-06,
+      "loss": 1.5485,
+      "step": 3655
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 1.520175498977463e-06,
+      "loss": 1.2581,
+      "step": 3656
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 1.514226265676011e-06,
+      "loss": 1.6864,
+      "step": 3657
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 1.5082882383657915e-06,
+      "loss": 1.6939,
+      "step": 3658
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 1.5023614206464077e-06,
+      "loss": 1.3633,
+      "step": 3659
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 1.4964458161106699e-06,
+      "loss": 1.629,
+      "step": 3660
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 1.4905414283445917e-06,
+      "loss": 1.8328,
+      "step": 3661
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 1.4846482609273839e-06,
+      "loss": 1.64,
+      "step": 3662
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 1.478766317431457e-06,
+      "loss": 1.4528,
+      "step": 3663
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 1.4728956014224216e-06,
+      "loss": 1.3892,
+      "step": 3664
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 1.4670361164590775e-06,
+      "loss": 1.4627,
+      "step": 3665
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 1.4611878660934187e-06,
+      "loss": 1.3123,
+      "step": 3666
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 1.455350853870623e-06,
+      "loss": 1.6183,
+      "step": 3667
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 1.4495250833290641e-06,
+      "loss": 1.5683,
+      "step": 3668
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 1.4437105580002952e-06,
+      "loss": 1.757,
+      "step": 3669
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 1.4379072814090566e-06,
+      "loss": 1.859,
+      "step": 3670
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 1.4321152570732698e-06,
+      "loss": 1.461,
+      "step": 3671
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 1.4263344885040286e-06,
+      "loss": 1.2332,
+      "step": 3672
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 1.4205649792056142e-06,
+      "loss": 1.7809,
+      "step": 3673
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 1.4148067326754711e-06,
+      "loss": 1.5907,
+      "step": 3674
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 1.409059752404225e-06,
+      "loss": 1.5091,
+      "step": 3675
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 1.4033240418756688e-06,
+      "loss": 1.6191,
+      "step": 3676
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 1.3975996045667616e-06,
+      "loss": 1.466,
+      "step": 3677
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 1.3918864439476342e-06,
+      "loss": 1.5655,
+      "step": 3678
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 1.3861845634815784e-06,
+      "loss": 1.9486,
+      "step": 3679
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 1.3804939666250473e-06,
+      "loss": 1.5819,
+      "step": 3680
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 1.3748146568276522e-06,
+      "loss": 1.3823,
+      "step": 3681
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 1.369146637532166e-06,
+      "loss": 1.47,
+      "step": 3682
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 1.3634899121745182e-06,
+      "loss": 2.0747,
+      "step": 3683
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 1.3578444841837858e-06,
+      "loss": 1.322,
+      "step": 3684
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 1.3522103569822042e-06,
+      "loss": 1.3019,
+      "step": 3685
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 1.346587533985153e-06,
+      "loss": 1.5254,
+      "step": 3686
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 1.3409760186011655e-06,
+      "loss": 1.8178,
+      "step": 3687
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 1.335375814231914e-06,
+      "loss": 1.8107,
+      "step": 3688
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 1.3297869242722183e-06,
+      "loss": 1.8866,
+      "step": 3689
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 1.3242093521100353e-06,
+      "loss": 1.4292,
+      "step": 3690
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 1.3186431011264666e-06,
+      "loss": 1.5969,
+      "step": 3691
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 1.3130881746957447e-06,
+      "loss": 1.5224,
+      "step": 3692
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 1.3075445761852468e-06,
+      "loss": 2.0205,
+      "step": 3693
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 1.3020123089554692e-06,
+      "loss": 1.481,
+      "step": 3694
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 1.2964913763600496e-06,
+      "loss": 1.5416,
+      "step": 3695
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 1.2909817817457527e-06,
+      "loss": 1.0432,
+      "step": 3696
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 1.285483528452469e-06,
+      "loss": 1.449,
+      "step": 3697
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 1.2799966198132153e-06,
+      "loss": 1.409,
+      "step": 3698
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 1.2745210591541302e-06,
+      "loss": 1.2134,
+      "step": 3699
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 1.2690568497944766e-06,
+      "loss": 1.5348,
+      "step": 3700
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 1.2636039950466273e-06,
+      "loss": 1.5301,
+      "step": 3701
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 1.2581624982160779e-06,
+      "loss": 1.4867,
+      "step": 3702
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 1.2527323626014431e-06,
+      "loss": 1.5907,
+      "step": 3703
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 1.2473135914944434e-06,
+      "loss": 1.7343,
+      "step": 3704
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 1.2419061881799133e-06,
+      "loss": 1.3738,
+      "step": 3705
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 1.2365101559357972e-06,
+      "loss": 1.5839,
+      "step": 3706
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 1.2311254980331477e-06,
+      "loss": 1.359,
+      "step": 3707
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 1.2257522177361137e-06,
+      "loss": 1.6212,
+      "step": 3708
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 1.220390318301954e-06,
+      "loss": 1.3796,
+      "step": 3709
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 1.2150398029810306e-06,
+      "loss": 1.4173,
+      "step": 3710
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 1.2097006750167984e-06,
+      "loss": 1.5547,
+      "step": 3711
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 1.2043729376458146e-06,
+      "loss": 1.6218,
+      "step": 3712
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 1.1990565940977294e-06,
+      "loss": 1.984,
+      "step": 3713
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 1.1937516475952826e-06,
+      "loss": 1.908,
+      "step": 3714
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 1.188458101354313e-06,
+      "loss": 1.3331,
+      "step": 3715
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 1.18317595858374e-06,
+      "loss": 1.1591,
+      "step": 3716
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 1.177905222485578e-06,
+      "loss": 1.2754,
+      "step": 3717
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 1.172645896254918e-06,
+      "loss": 1.6976,
+      "step": 3718
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 1.1673979830799453e-06,
+      "loss": 1.6439,
+      "step": 3719
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 1.1621614861419194e-06,
+      "loss": 1.4912,
+      "step": 3720
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 1.1569364086151768e-06,
+      "loss": 1.3587,
+      "step": 3721
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 1.151722753667137e-06,
+      "loss": 1.7894,
+      "step": 3722
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 1.1465205244582966e-06,
+      "loss": 1.4789,
+      "step": 3723
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 1.1413297241422194e-06,
+      "loss": 1.5418,
+      "step": 3724
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 1.1361503558655485e-06,
+      "loss": 1.4515,
+      "step": 3725
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 1.13098242276799e-06,
+      "loss": 1.4703,
+      "step": 3726
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 1.1258259279823225e-06,
+      "loss": 1.5319,
+      "step": 3727
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 1.1206808746343879e-06,
+      "loss": 1.3467,
+      "step": 3728
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 1.1155472658430976e-06,
+      "loss": 1.6609,
+      "step": 3729
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 1.1104251047204185e-06,
+      "loss": 1.3676,
+      "step": 3730
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 1.105314394371384e-06,
+      "loss": 1.5306,
+      "step": 3731
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 1.1002151378940785e-06,
+      "loss": 1.6625,
+      "step": 3732
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 1.0951273383796535e-06,
+      "loss": 1.9965,
+      "step": 3733
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 1.0900509989123088e-06,
+      "loss": 1.294,
+      "step": 3734
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 1.0849861225692936e-06,
+      "loss": 1.5093,
+      "step": 3735
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 1.0799327124209168e-06,
+      "loss": 1.475,
+      "step": 3736
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 1.0748907715305301e-06,
+      "loss": 1.2642,
+      "step": 3737
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 1.069860302954535e-06,
+      "loss": 1.2895,
+      "step": 3738
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 1.0648413097423816e-06,
+      "loss": 1.5356,
+      "step": 3739
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 1.0598337949365577e-06,
+      "loss": 1.3689,
+      "step": 3740
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 1.0548377615725957e-06,
+      "loss": 1.6641,
+      "step": 3741
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 1.0498532126790684e-06,
+      "loss": 1.602,
+      "step": 3742
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 1.0448801512775897e-06,
+      "loss": 1.4403,
+      "step": 3743
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 1.0399185803828016e-06,
+      "loss": 1.6044,
+      "step": 3744
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 1.0349685030023893e-06,
+      "loss": 2.258,
+      "step": 3745
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 1.0300299221370635e-06,
+      "loss": 1.5843,
+      "step": 3746
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 1.0251028407805742e-06,
+      "loss": 1.7997,
+      "step": 3747
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 1.020187261919694e-06,
+      "loss": 1.8671,
+      "step": 3748
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 1.0152831885342195e-06,
+      "loss": 1.4349,
+      "step": 3749
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 1.010390623596982e-06,
+      "loss": 1.7222,
+      "step": 3750
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 1.0055095700738305e-06,
+      "loss": 1.7083,
+      "step": 3751
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 1.0006400309236385e-06,
+      "loss": 1.5454,
+      "step": 3752
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 9.957820090982962e-07,
+      "loss": 1.6384,
+      "step": 3753
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 9.909355075427163e-07,
+      "loss": 1.5436,
+      "step": 3754
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 9.861005291948244e-07,
+      "loss": 1.1726,
+      "step": 3755
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 9.81277076985563e-07,
+      "loss": 1.3178,
+      "step": 3756
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 9.764651538388858e-07,
+      "loss": 1.4604,
+      "step": 3757
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 9.7166476267176e-07,
+      "loss": 1.5917,
+      "step": 3758
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 9.668759063941603e-07,
+      "loss": 1.4172,
+      "step": 3759
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 9.620985879090683e-07,
+      "loss": 1.4236,
+      "step": 3760
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 9.573328101124747e-07,
+      "loss": 1.3154,
+      "step": 3761
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 9.525785758933681e-07,
+      "loss": 1.5958,
+      "step": 3762
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 9.478358881337479e-07,
+      "loss": 2.0128,
+      "step": 3763
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 9.431047497086054e-07,
+      "loss": 1.475,
+      "step": 3764
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 9.383851634859398e-07,
+      "loss": 1.7278,
+      "step": 3765
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 9.33677132326738e-07,
+      "loss": 1.4826,
+      "step": 3766
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 9.289806590849904e-07,
+      "loss": 1.8617,
+      "step": 3767
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 9.242957466076752e-07,
+      "loss": 1.2947,
+      "step": 3768
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 9.196223977347696e-07,
+      "loss": 1.5291,
+      "step": 3769
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 9.149606152992296e-07,
+      "loss": 1.7631,
+      "step": 3770
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 9.103104021270082e-07,
+      "loss": 1.8117,
+      "step": 3771
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 9.056717610370435e-07,
+      "loss": 1.8309,
+      "step": 3772
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 9.010446948412621e-07,
+      "loss": 1.7314,
+      "step": 3773
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 8.964292063445645e-07,
+      "loss": 1.8407,
+      "step": 3774
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 8.918252983448439e-07,
+      "loss": 1.4202,
+      "step": 3775
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 8.872329736329654e-07,
+      "loss": 1.5746,
+      "step": 3776
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 8.82652234992778e-07,
+      "loss": 1.198,
+      "step": 3777
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 8.780830852011002e-07,
+      "loss": 1.6083,
+      "step": 3778
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 8.735255270277299e-07,
+      "loss": 1.8101,
+      "step": 3779
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 8.689795632354414e-07,
+      "loss": 1.7768,
+      "step": 3780
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 8.64445196579975e-07,
+      "loss": 1.3721,
+      "step": 3781
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 8.599224298100428e-07,
+      "loss": 1.6314,
+      "step": 3782
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 8.554112656673253e-07,
+      "loss": 1.5264,
+      "step": 3783
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 8.509117068864703e-07,
+      "loss": 1.8407,
+      "step": 3784
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 8.464237561950894e-07,
+      "loss": 1.4479,
+      "step": 3785
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 8.419474163137575e-07,
+      "loss": 1.4303,
+      "step": 3786
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 8.374826899560129e-07,
+      "loss": 1.7592,
+      "step": 3787
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 8.330295798283527e-07,
+      "loss": 1.3484,
+      "step": 3788
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 8.285880886302267e-07,
+      "loss": 1.4005,
+      "step": 3789
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 8.241582190540498e-07,
+      "loss": 1.6065,
+      "step": 3790
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 8.197399737851875e-07,
+      "loss": 1.2884,
+      "step": 3791
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 8.153333555019638e-07,
+      "loss": 1.541,
+      "step": 3792
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 8.109383668756443e-07,
+      "loss": 1.7575,
+      "step": 3793
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 8.065550105704556e-07,
+      "loss": 1.507,
+      "step": 3794
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 8.021832892435677e-07,
+      "loss": 1.8598,
+      "step": 3795
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 7.978232055450963e-07,
+      "loss": 1.6075,
+      "step": 3796
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 7.934747621181049e-07,
+      "loss": 1.6133,
+      "step": 3797
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 7.891379615986006e-07,
+      "loss": 1.0964,
+      "step": 3798
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 7.848128066155314e-07,
+      "loss": 1.3503,
+      "step": 3799
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 7.804992997907845e-07,
+      "loss": 1.4002,
+      "step": 3800
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 7.761974437391906e-07,
+      "loss": 1.3189,
+      "step": 3801
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 7.719072410685169e-07,
+      "loss": 1.2708,
+      "step": 3802
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 7.676286943794609e-07,
+      "loss": 1.1956,
+      "step": 3803
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 7.633618062656589e-07,
+      "loss": 1.2857,
+      "step": 3804
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 7.591065793136776e-07,
+      "loss": 1.5659,
+      "step": 3805
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 7.548630161030179e-07,
+      "loss": 1.5343,
+      "step": 3806
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 7.506311192061088e-07,
+      "loss": 1.6694,
+      "step": 3807
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 7.46410891188305e-07,
+      "loss": 1.3995,
+      "step": 3808
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 7.42202334607891e-07,
+      "loss": 1.1903,
+      "step": 3809
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 7.380054520160729e-07,
+      "loss": 1.3462,
+      "step": 3810
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 7.338202459569843e-07,
+      "loss": 1.5222,
+      "step": 3811
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 7.296467189676737e-07,
+      "loss": 1.6297,
+      "step": 3812
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 7.254848735781195e-07,
+      "loss": 1.5068,
+      "step": 3813
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 7.21334712311208e-07,
+      "loss": 1.2175,
+      "step": 3814
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 7.171962376827534e-07,
+      "loss": 1.4986,
+      "step": 3815
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 7.130694522014736e-07,
+      "loss": 1.4654,
+      "step": 3816
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 7.089543583690117e-07,
+      "loss": 1.3267,
+      "step": 3817
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 7.048509586799168e-07,
+      "loss": 1.668,
+      "step": 3818
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 7.007592556216502e-07,
+      "loss": 1.4349,
+      "step": 3819
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 6.966792516745858e-07,
+      "loss": 1.5675,
+      "step": 3820
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 6.926109493120025e-07,
+      "loss": 1.7105,
+      "step": 3821
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 6.885543510000858e-07,
+      "loss": 1.5463,
+      "step": 3822
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 6.845094591979284e-07,
+      "loss": 1.2849,
+      "step": 3823
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 6.804762763575246e-07,
+      "loss": 1.8341,
+      "step": 3824
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 6.764548049237695e-07,
+      "loss": 1.6418,
+      "step": 3825
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 6.724450473344623e-07,
+      "loss": 1.8697,
+      "step": 3826
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 6.684470060202986e-07,
+      "loss": 1.6277,
+      "step": 3827
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 6.64460683404875e-07,
+      "loss": 1.6236,
+      "step": 3828
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 6.604860819046833e-07,
+      "loss": 1.6831,
+      "step": 3829
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 6.565232039291025e-07,
+      "loss": 1.3268,
+      "step": 3830
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 6.525720518804135e-07,
+      "loss": 1.8198,
+      "step": 3831
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 6.486326281537891e-07,
+      "loss": 1.5092,
+      "step": 3832
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 6.44704935137288e-07,
+      "loss": 1.11,
+      "step": 3833
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 6.407889752118613e-07,
+      "loss": 1.695,
+      "step": 3834
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 6.368847507513432e-07,
+      "loss": 1.4818,
+      "step": 3835
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 6.329922641224562e-07,
+      "loss": 1.297,
+      "step": 3836
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 6.291115176848128e-07,
+      "loss": 1.5983,
+      "step": 3837
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 6.252425137909002e-07,
+      "loss": 1.9303,
+      "step": 3838
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 6.21385254786091e-07,
+      "loss": 1.6189,
+      "step": 3839
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 6.175397430086372e-07,
+      "loss": 1.6047,
+      "step": 3840
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 6.137059807896717e-07,
+      "loss": 1.5787,
+      "step": 3841
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 6.098839704532045e-07,
+      "loss": 1.2676,
+      "step": 3842
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 6.060737143161177e-07,
+      "loss": 1.3569,
+      "step": 3843
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 6.022752146881728e-07,
+      "loss": 1.5893,
+      "step": 3844
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 5.984884738720031e-07,
+      "loss": 1.5905,
+      "step": 3845
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 5.947134941631172e-07,
+      "loss": 1.4544,
+      "step": 3846
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 5.909502778498821e-07,
+      "loss": 1.743,
+      "step": 3847
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 5.871988272135487e-07,
+      "loss": 1.5662,
+      "step": 3848
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 5.834591445282245e-07,
+      "loss": 1.6365,
+      "step": 3849
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 5.797312320608939e-07,
+      "loss": 1.4951,
+      "step": 3850
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 5.760150920713959e-07,
+      "loss": 1.5093,
+      "step": 3851
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 5.723107268124372e-07,
+      "loss": 1.6801,
+      "step": 3852
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 5.686181385295886e-07,
+      "loss": 1.4797,
+      "step": 3853
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 5.649373294612814e-07,
+      "loss": 1.4902,
+      "step": 3854
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 5.612683018388044e-07,
+      "loss": 1.3247,
+      "step": 3855
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 5.57611057886307e-07,
+      "loss": 1.5073,
+      "step": 3856
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 5.539655998207872e-07,
+      "loss": 1.419,
+      "step": 3857
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 5.503319298521104e-07,
+      "loss": 1.522,
+      "step": 3858
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 5.467100501829903e-07,
+      "loss": 1.852,
+      "step": 3859
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 5.430999630089929e-07,
+      "loss": 1.6106,
+      "step": 3860
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 5.395016705185363e-07,
+      "loss": 2.0535,
+      "step": 3861
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 5.359151748928892e-07,
+      "loss": 1.8767,
+      "step": 3862
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 5.323404783061681e-07,
+      "loss": 1.607,
+      "step": 3863
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 5.287775829253395e-07,
+      "loss": 1.2409,
+      "step": 3864
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 5.252264909102134e-07,
+      "loss": 1.9762,
+      "step": 3865
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 5.216872044134435e-07,
+      "loss": 1.4031,
+      "step": 3866
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 5.181597255805315e-07,
+      "loss": 1.7489,
+      "step": 3867
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 5.146440565498178e-07,
+      "loss": 1.5775,
+      "step": 3868
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 5.111401994524867e-07,
+      "loss": 1.9865,
+      "step": 3869
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 5.076481564125568e-07,
+      "loss": 1.7606,
+      "step": 3870
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 5.041679295468926e-07,
+      "loss": 1.8119,
+      "step": 3871
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 5.006995209651866e-07,
+      "loss": 1.8968,
+      "step": 3872
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 4.972429327699768e-07,
+      "loss": 1.6249,
+      "step": 3873
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 4.937981670566294e-07,
+      "loss": 1.4366,
+      "step": 3874
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 4.903652259133451e-07,
+      "loss": 1.5635,
+      "step": 3875
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 4.86944111421157e-07,
+      "loss": 1.299,
+      "step": 3876
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 4.835348256539307e-07,
+      "loss": 1.5373,
+      "step": 3877
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 4.801373706783597e-07,
+      "loss": 1.3091,
+      "step": 3878
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 4.767517485539608e-07,
+      "loss": 1.4146,
+      "step": 3879
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 4.7337796133309024e-07,
+      "loss": 1.3959,
+      "step": 3880
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 4.700160110609142e-07,
+      "loss": 1.5698,
+      "step": 3881
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 4.666658997754381e-07,
+      "loss": 1.5572,
+      "step": 3882
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 4.63327629507484e-07,
+      "loss": 1.5378,
+      "step": 3883
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 4.60001202280691e-07,
+      "loss": 1.1778,
+      "step": 3884
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 4.566866201115261e-07,
+      "loss": 1.7379,
+      "step": 3885
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 4.533838850092731e-07,
+      "loss": 1.5248,
+      "step": 3886
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 4.500929989760372e-07,
+      "loss": 1.5827,
+      "step": 3887
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 4.4681396400673814e-07,
+      "loss": 1.9115,
+      "step": 3888
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 4.435467820891104e-07,
+      "loss": 1.5828,
+      "step": 3889
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 4.402914552037052e-07,
+      "loss": 1.9609,
+      "step": 3890
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 4.3704798532388624e-07,
+      "loss": 1.3312,
+      "step": 3891
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 4.338163744158319e-07,
+      "loss": 1.6616,
+      "step": 3892
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 4.305966244385285e-07,
+      "loss": 1.8841,
+      "step": 3893
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 4.273887373437746e-07,
+      "loss": 1.713,
+      "step": 3894
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 4.241927150761749e-07,
+      "loss": 1.5218,
+      "step": 3895
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 4.210085595731461e-07,
+      "loss": 1.3204,
+      "step": 3896
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 4.1783627276490433e-07,
+      "loss": 1.322,
+      "step": 3897
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 4.14675856574478e-07,
+      "loss": 1.2805,
+      "step": 3898
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 4.1152731291769886e-07,
+      "loss": 1.6465,
+      "step": 3899
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 4.083906437031959e-07,
+      "loss": 1.7734,
+      "step": 3900
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 4.0526585083240586e-07,
+      "loss": 1.8074,
+      "step": 3901
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 4.0215293619956466e-07,
+      "loss": 1.3187,
+      "step": 3902
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 3.990519016917049e-07,
+      "loss": 1.3325,
+      "step": 3903
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 3.959627491886586e-07,
+      "loss": 1.6575,
+      "step": 3904
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 3.928854805630589e-07,
+      "loss": 1.7215,
+      "step": 3905
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 3.8982009768033127e-07,
+      "loss": 1.6362,
+      "step": 3906
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 3.8676660239869377e-07,
+      "loss": 1.2608,
+      "step": 3907
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 3.837249965691658e-07,
+      "loss": 1.3355,
+      "step": 3908
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 3.806952820355503e-07,
+      "loss": 1.6514,
+      "step": 3909
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 3.776774606344491e-07,
+      "loss": 1.4577,
+      "step": 3910
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 3.746715341952523e-07,
+      "loss": 1.5835,
+      "step": 3911
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 3.716775045401355e-07,
+      "loss": 1.7664,
+      "step": 3912
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 3.6869537348406706e-07,
+      "loss": 1.4002,
+      "step": 3913
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 3.65725142834803e-07,
+      "loss": 1.7067,
+      "step": 3914
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 3.627668143928853e-07,
+      "loss": 1.4857,
+      "step": 3915
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 3.598203899516328e-07,
+      "loss": 1.3648,
+      "step": 3916
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 3.568858712971568e-07,
+      "loss": 1.3116,
+      "step": 3917
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 3.5396326020835205e-07,
+      "loss": 1.6891,
+      "step": 3918
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 3.51052558456888e-07,
+      "loss": 1.3935,
+      "step": 3919
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 3.481537678072222e-07,
+      "loss": 1.7,
+      "step": 3920
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 3.4526689001658455e-07,
+      "loss": 1.2706,
+      "step": 3921
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 3.4239192683499066e-07,
+      "loss": 1.7196,
+      "step": 3922
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 3.3952888000522877e-07,
+      "loss": 1.6144,
+      "step": 3923
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 3.3667775126286605e-07,
+      "loss": 1.3852,
+      "step": 3924
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 3.3383854233624003e-07,
+      "loss": 1.4442,
+      "step": 3925
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 3.310112549464695e-07,
+      "loss": 1.7261,
+      "step": 3926
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 3.281958908074412e-07,
+      "loss": 1.6194,
+      "step": 3927
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 3.2539245162581666e-07,
+      "loss": 1.3911,
+      "step": 3928
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 3.2260093910102764e-07,
+      "loss": 1.5012,
+      "step": 3929
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 3.19821354925276e-07,
+      "loss": 1.5604,
+      "step": 3930
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 3.1705370078353393e-07,
+      "loss": 1.9402,
+      "step": 3931
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 3.1429797835353935e-07,
+      "loss": 1.5631,
+      "step": 3932
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 3.115541893057983e-07,
+      "loss": 1.9806,
+      "step": 3933
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 3.0882233530358686e-07,
+      "loss": 1.3141,
+      "step": 3934
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 3.0610241800293815e-07,
+      "loss": 1.9015,
+      "step": 3935
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 3.033944390526555e-07,
+      "loss": 1.5115,
+      "step": 3936
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 3.006984000943036e-07,
+      "loss": 1.6901,
+      "step": 3937
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 2.9801430276220625e-07,
+      "loss": 1.8221,
+      "step": 3938
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 2.953421486834551e-07,
+      "loss": 1.5741,
+      "step": 3939
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 2.926819394778946e-07,
+      "loss": 1.5123,
+      "step": 3940
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 2.900336767581302e-07,
+      "loss": 1.2131,
+      "step": 3941
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 2.873973621295312e-07,
+      "loss": 1.5981,
+      "step": 3942
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 2.847729971902147e-07,
+      "loss": 1.3817,
+      "step": 3943
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 2.8216058353106147e-07,
+      "loss": 1.4318,
+      "step": 3944
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 2.795601227357048e-07,
+      "loss": 1.7387,
+      "step": 3945
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 2.7697161638053025e-07,
+      "loss": 1.5506,
+      "step": 3946
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 2.7439506603468056e-07,
+      "loss": 1.5632,
+      "step": 3947
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 2.718304732600485e-07,
+      "loss": 1.1902,
+      "step": 3948
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 2.692778396112772e-07,
+      "loss": 1.4257,
+      "step": 3949
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 2.6673716663575986e-07,
+      "loss": 1.8572,
+      "step": 3950
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 2.6420845587364684e-07,
+      "loss": 1.7859,
+      "step": 3951
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 2.6169170885782527e-07,
+      "loss": 1.946,
+      "step": 3952
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 2.591869271139369e-07,
+      "loss": 1.4061,
+      "step": 3953
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 2.5669411216036943e-07,
+      "loss": 1.6153,
+      "step": 3954
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 2.542132655082541e-07,
+      "loss": 1.8186,
+      "step": 3955
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 2.517443886614723e-07,
+      "loss": 1.9032,
+      "step": 3956
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 2.4928748311664475e-07,
+      "loss": 1.7024,
+      "step": 3957
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 2.468425503631355e-07,
+      "loss": 1.413,
+      "step": 3958
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 2.4440959188305027e-07,
+      "loss": 1.6145,
+      "step": 3959
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 2.419886091512402e-07,
+      "loss": 1.5958,
+      "step": 3960
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 2.3957960363529374e-07,
+      "loss": 1.4124,
+      "step": 3961
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 2.3718257679553603e-07,
+      "loss": 1.5047,
+      "step": 3962
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 2.347975300850358e-07,
+      "loss": 1.3686,
+      "step": 3963
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 2.3242446494959657e-07,
+      "loss": 1.8043,
+      "step": 3964
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 2.3006338282776099e-07,
+      "loss": 1.2865,
+      "step": 3965
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 2.2771428515080408e-07,
+      "loss": 1.1027,
+      "step": 3966
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 2.2537717334274012e-07,
+      "loss": 1.6436,
+      "step": 3967
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 2.2305204882031583e-07,
+      "loss": 1.398,
+      "step": 3968
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 2.207389129930082e-07,
+      "loss": 1.5645,
+      "step": 3969
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 2.1843776726303112e-07,
+      "loss": 1.5776,
+      "step": 3970
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 2.1614861302532876e-07,
+      "loss": 1.3836,
+      "step": 3971
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 2.1387145166757773e-07,
+      "loss": 1.7867,
+      "step": 3972
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 2.116062845701783e-07,
+      "loss": 1.1956,
+      "step": 3973
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 2.0935311310626983e-07,
+      "loss": 1.3491,
+      "step": 3974
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 2.0711193864170864e-07,
+      "loss": 1.851,
+      "step": 3975
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 2.048827625350902e-07,
+      "loss": 1.546,
+      "step": 3976
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 2.026655861377269e-07,
+      "loss": 1.4186,
+      "step": 3977
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 2.0046041079366364e-07,
+      "loss": 1.6882,
+      "step": 3978
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 1.982672378396644e-07,
+      "loss": 1.9181,
+      "step": 3979
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 1.960860686052235e-07,
+      "loss": 1.3128,
+      "step": 3980
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 1.9391690441255219e-07,
+      "loss": 1.8941,
+      "step": 3981
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 1.9175974657659192e-07,
+      "loss": 1.235,
+      "step": 3982
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 1.8961459640499892e-07,
+      "loss": 1.6414,
+      "step": 3983
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 1.8748145519815742e-07,
+      "loss": 1.7996,
+      "step": 3984
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 1.8536032424916194e-07,
+      "loss": 1.6823,
+      "step": 3985
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 1.832512048438373e-07,
+      "loss": 1.8836,
+      "step": 3986
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 1.811540982607185e-07,
+      "loss": 1.659,
+      "step": 3987
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 1.7906900577106422e-07,
+      "loss": 1.5635,
+      "step": 3988
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 1.7699592863884785e-07,
+      "loss": 1.7113,
+      "step": 3989
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 1.7493486812075743e-07,
+      "loss": 1.3637,
+      "step": 3990
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 1.7288582546620024e-07,
+      "loss": 1.4822,
+      "step": 3991
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 1.7084880191729601e-07,
+      "loss": 1.7053,
+      "step": 3992
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 1.688237987088792e-07,
+      "loss": 1.6334,
+      "step": 3993
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 1.6681081706849677e-07,
+      "loss": 1.2675,
+      "step": 3994
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 1.6480985821640815e-07,
+      "loss": 1.8372,
+      "step": 3995
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 1.6282092336558975e-07,
+      "loss": 1.2304,
+      "step": 3996
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 1.6084401372171932e-07,
+      "loss": 1.3083,
+      "step": 3997
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 1.5887913048319158e-07,
+      "loss": 1.4168,
+      "step": 3998
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 1.5692627484111377e-07,
+      "loss": 1.5042,
+      "step": 3999
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 1.5498544797929226e-07,
+      "loss": 1.2526,
+      "step": 4000
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 1.530566510742526e-07,
+      "loss": 1.8032,
+      "step": 4001
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 1.511398852952195e-07,
+      "loss": 1.4906,
+      "step": 4002
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 1.4923515180412794e-07,
+      "loss": 1.3018,
+      "step": 4003
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 1.47342451755621e-07,
+      "loss": 1.7271,
+      "step": 4004
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 1.454617862970431e-07,
+      "loss": 1.284,
+      "step": 4005
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 1.435931565684423e-07,
+      "loss": 1.5071,
+      "step": 4006
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 1.4173656370257693e-07,
+      "loss": 1.3975,
+      "step": 4007
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 1.3989200882490227e-07,
+      "loss": 1.3491,
+      "step": 4008
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 1.380594930535839e-07,
+      "loss": 1.3285,
+      "step": 4009
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 1.3623901749947989e-07,
+      "loss": 1.5085,
+      "step": 4010
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 1.344305832661541e-07,
+      "loss": 1.6516,
+      "step": 4011
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 1.326341914498741e-07,
+      "loss": 1.6279,
+      "step": 4012
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 1.3084984313960214e-07,
+      "loss": 1.4246,
+      "step": 4013
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 1.2907753941700407e-07,
+      "loss": 1.3516,
+      "step": 4014
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 1.2731728135643828e-07,
+      "loss": 1.2295,
+      "step": 4015
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 1.25569070024969e-07,
+      "loss": 1.6022,
+      "step": 4016
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 1.2383290648235292e-07,
+      "loss": 1.6948,
+      "step": 4017
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 1.2210879178104597e-07,
+      "loss": 1.5851,
+      "step": 4018
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 1.203967269661943e-07,
+      "loss": 1.903,
+      "step": 4019
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 1.1869671307564778e-07,
+      "loss": 1.8133,
+      "step": 4020
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 1.1700875113994425e-07,
+      "loss": 1.84,
+      "step": 4021
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 1.1533284218231855e-07,
+      "loss": 1.731,
+      "step": 4022
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 1.1366898721870245e-07,
+      "loss": 1.413,
+      "step": 4023
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 1.120171872577136e-07,
+      "loss": 1.5756,
+      "step": 4024
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 1.1037744330066658e-07,
+      "loss": 1.6721,
+      "step": 4025
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 1.0874975634156625e-07,
+      "loss": 1.2701,
+      "step": 4026
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 1.0713412736711227e-07,
+      "loss": 1.182,
+      "step": 4027
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 1.055305573566856e-07,
+      "loss": 1.4051,
+      "step": 4028
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 1.039390472823687e-07,
+      "loss": 1.4586,
+      "step": 4029
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 1.0235959810892538e-07,
+      "loss": 1.2806,
+      "step": 4030
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 1.0079221079380752e-07,
+      "loss": 1.8304,
+      "step": 4031
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 9.9236886287164e-08,
+      "loss": 1.4832,
+      "step": 4032
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 9.769362553182282e-08,
+      "loss": 1.7305,
+      "step": 4033
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 9.61624294632979e-08,
+      "loss": 1.6234,
+      "step": 4034
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 9.464329900980007e-08,
+      "loss": 1.7392,
+      "step": 4035
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 9.31362350922127e-08,
+      "loss": 1.4322,
+      "step": 4036
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 9.164123862411612e-08,
+      "loss": 1.4648,
+      "step": 4037
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 9.015831051176537e-08,
+      "loss": 1.5891,
+      "step": 4038
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 8.86874516541103e-08,
+      "loss": 1.2284,
+      "step": 4039
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 8.722866294277543e-08,
+      "loss": 1.3866,
+      "step": 4040
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 8.578194526207118e-08,
+      "loss": 1.3902,
+      "step": 4041
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 8.434729948899378e-08,
+      "loss": 1.8258,
+      "step": 4042
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 8.292472649321648e-08,
+      "loss": 1.4882,
+      "step": 4043
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 8.151422713709611e-08,
+      "loss": 1.1618,
+      "step": 4044
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 8.011580227567539e-08,
+      "loss": 1.5519,
+      "step": 4045
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 7.872945275666954e-08,
+      "loss": 1.2248,
+      "step": 4046
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 7.735517942047521e-08,
+      "loss": 1.8723,
+      "step": 4047
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 7.599298310017488e-08,
+      "loss": 1.8598,
+      "step": 4048
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 7.464286462152359e-08,
+      "loss": 1.8018,
+      "step": 4049
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 7.330482480295775e-08,
+      "loss": 1.9624,
+      "step": 4050
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 7.197886445559077e-08,
+      "loss": 1.6449,
+      "step": 4051
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 7.0664984383213e-08,
+      "loss": 1.6453,
+      "step": 4052
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 6.936318538229181e-08,
+      "loss": 1.3842,
+      "step": 4053
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 6.80734682419737e-08,
+      "loss": 2.2304,
+      "step": 4054
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 6.679583374407994e-08,
+      "loss": 1.3769,
+      "step": 4055
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 6.553028266310434e-08,
+      "loss": 1.433,
+      "step": 4056
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 6.42768157662177e-08,
+      "loss": 1.9031,
+      "step": 4057
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 6.303543381326993e-08,
+      "loss": 1.3139,
+      "step": 4058
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 6.18061375567769e-08,
+      "loss": 1.6585,
+      "step": 4059
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 6.058892774193359e-08,
+      "loss": 1.2516,
+      "step": 4060
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 5.938380510660757e-08,
+      "loss": 1.7237,
+      "step": 4061
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 5.819077038133891e-08,
+      "loss": 1.7597,
+      "step": 4062
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 5.700982428934021e-08,
+      "loss": 1.3313,
+      "step": 4063
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 5.5840967546492154e-08,
+      "loss": 1.7921,
+      "step": 4064
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 5.468420086135462e-08,
+      "loss": 1.5268,
+      "step": 4065
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 5.353952493515113e-08,
+      "loss": 1.3219,
+      "step": 4066
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 5.240694046178219e-08,
+      "loss": 1.4245,
+      "step": 4067
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 5.1286448127811916e-08,
+      "loss": 1.527,
+      "step": 4068
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 5.017804861247921e-08,
+      "loss": 1.1458,
+      "step": 4069
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 4.908174258769105e-08,
+      "loss": 1.6686,
+      "step": 4070
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 4.799753071802471e-08,
+      "loss": 1.4533,
+      "step": 4071
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 4.692541366072112e-08,
+      "loss": 1.6468,
+      "step": 4072
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 4.586539206569374e-08,
+      "loss": 1.4102,
+      "step": 4073
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 4.481746657552188e-08,
+      "loss": 1.3814,
+      "step": 4074
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 4.378163782545741e-08,
+      "loss": 1.1626,
+      "step": 4075
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 4.2757906443409156e-08,
+      "loss": 1.2065,
+      "step": 4076
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 4.1746273049962926e-08,
+      "loss": 1.9526,
+      "step": 4077
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 4.074673825836373e-08,
+      "loss": 1.6455,
+      "step": 4078
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 3.975930267452688e-08,
+      "loss": 1.7455,
+      "step": 4079
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 3.878396689702912e-08,
+      "loss": 1.7859,
+      "step": 4080
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 3.7820731517117514e-08,
+      "loss": 1.5451,
+      "step": 4081
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 3.6869597118698306e-08,
+      "loss": 1.4857,
+      "step": 4082
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 3.593056427834363e-08,
+      "loss": 1.4321,
+      "step": 4083
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 3.5003633565295905e-08,
+      "loss": 1.3652,
+      "step": 4084
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.408880554145011e-08,
+      "loss": 2.071,
+      "step": 4085
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3.318608076137597e-08,
+      "loss": 1.8416,
+      "step": 4086
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 3.2295459772297976e-08,
+      "loss": 1.6851,
+      "step": 4087
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 3.14169431141087e-08,
+      "loss": 1.2844,
+      "step": 4088
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 3.055053131935992e-08,
+      "loss": 1.6838,
+      "step": 4089
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 2.9696224913264848e-08,
+      "loss": 1.2436,
+      "step": 4090
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 2.8854024413702554e-08,
+      "loss": 1.6556,
+      "step": 4091
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 2.8023930331209094e-08,
+      "loss": 1.5784,
+      "step": 4092
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 2.7205943168984172e-08,
+      "loss": 1.4986,
+      "step": 4093
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 2.6400063422888923e-08,
+      "loss": 1.4379,
+      "step": 4094
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 2.5606291581445897e-08,
+      "loss": 1.5209,
+      "step": 4095
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 2.4824628125830198e-08,
+      "loss": 1.0637,
+      "step": 4096
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 2.4055073529887228e-08,
+      "loss": 1.7054,
+      "step": 4097
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 2.3297628260117166e-08,
+      "loss": 1.6272,
+      "step": 4098
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 2.2552292775677164e-08,
+      "loss": 1.3852,
+      "step": 4099
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 2.1819067528388026e-08,
+      "loss": 1.5673,
+      "step": 4100
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 2.1097952962729762e-08,
+      "loss": 1.3995,
+      "step": 4101
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 2.03889495158327e-08,
+      "loss": 1.4144,
+      "step": 4102
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 1.9692057617497484e-08,
+      "loss": 1.4566,
+      "step": 4103
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 1.9007277690172853e-08,
+      "loss": 1.7567,
+      "step": 4104
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 1.8334610148973418e-08,
+      "loss": 1.5268,
+      "step": 4105
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 1.7674055401661893e-08,
+      "loss": 1.76,
+      "step": 4106
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 1.7025613848666856e-08,
+      "loss": 1.6064,
+      "step": 4107
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 1.6389285883069428e-08,
+      "loss": 1.3384,
+      "step": 4108
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 1.576507189060994e-08,
+      "loss": 1.3683,
+      "step": 4109
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 1.5152972249683483e-08,
+      "loss": 1.3446,
+      "step": 4110
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 1.4552987331342139e-08,
+      "loss": 1.6094,
+      "step": 4111
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 1.3965117499292746e-08,
+      "loss": 1.6584,
+      "step": 4112
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 1.3389363109899133e-08,
+      "loss": 1.5225,
+      "step": 4113
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 1.2825724512184334e-08,
+      "loss": 1.2579,
+      "step": 4114
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 1.2274202047821705e-08,
+      "loss": 1.706,
+      "step": 4115
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 1.1734796051139363e-08,
+      "loss": 1.4544,
+      "step": 4116
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 1.1207506849124639e-08,
+      "loss": 1.3734,
+      "step": 4117
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 1.0692334761419621e-08,
+      "loss": 1.9373,
+      "step": 4118
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 1.0189280100314502e-08,
+      "loss": 1.6197,
+      "step": 4119
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 9.698343170763125e-09,
+      "loss": 1.4032,
+      "step": 4120
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 9.21952427036965e-09,
+      "loss": 1.6162,
+      "step": 4121
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 8.752823689386347e-09,
+      "loss": 1.7136,
+      "step": 4122
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 8.298241710731347e-09,
+      "loss": 1.1391,
+      "step": 4123
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 7.855778609966446e-09,
+      "loss": 1.3474,
+      "step": 4124
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 7.425434655312647e-09,
+      "loss": 1.6151,
+      "step": 4125
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 7.0072101076390505e-09,
+      "loss": 1.8672,
+      "step": 4126
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 6.60110522047619e-09,
+      "loss": 1.2401,
+      "step": 4127
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 6.207120240000475e-09,
+      "loss": 1.7832,
+      "step": 4128
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 5.825255405043084e-09,
+      "loss": 1.4376,
+      "step": 4129
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 5.455510947089959e-09,
+      "loss": 1.44,
+      "step": 4130
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 5.0978870902773645e-09,
+      "loss": 1.6211,
+      "step": 4131
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 4.752384051398551e-09,
+      "loss": 1.5061,
+      "step": 4132
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 4.41900203989043e-09,
+      "loss": 1.9579,
+      "step": 4133
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 4.09774125785356e-09,
+      "loss": 1.5559,
+      "step": 4134
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 3.788601900029942e-09,
+      "loss": 1.66,
+      "step": 4135
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 3.4915841538207817e-09,
+      "loss": 1.8768,
+      "step": 4136
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 3.2066881992753874e-09,
+      "loss": 1.4144,
+      "step": 4137
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 2.933914209100053e-09,
+      "loss": 1.3436,
+      "step": 4138
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 2.673262348644734e-09,
+      "loss": 1.7806,
+      "step": 4139
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 2.424732775916372e-09,
+      "loss": 1.7769,
+      "step": 4140
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 2.1883256415744515e-09,
+      "loss": 1.1516,
+      "step": 4141
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 1.9640410889265606e-09,
+      "loss": 1.7068,
+      "step": 4142
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 1.7518792539328312e-09,
+      "loss": 1.5924,
+      "step": 4143
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 1.5518402652059394e-09,
+      "loss": 1.7384,
+      "step": 4144
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 1.3639242440066646e-09,
+      "loss": 1.7302,
+      "step": 4145
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 1.18813130424833e-09,
+      "loss": 1.614,
+      "step": 4146
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 1.0244615524990231e-09,
+      "loss": 1.7238,
+      "step": 4147
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 8.729150879749349e-10,
+      "loss": 2.3379,
+      "step": 4148
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 7.334920025381387e-10,
+      "loss": 1.606,
+      "step": 4149
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 6.061923807099135e-10,
+      "loss": 1.6865,
+      "step": 4150
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 4.910162996574209e-10,
+      "loss": 1.3075,
+      "step": 4151
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 3.8796382920036623e-10,
+      "loss": 1.7003,
+      "step": 4152
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 2.9703503180655845e-10,
+      "loss": 1.4335,
+      "step": 4153
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 2.182299626007911e-10,
+      "loss": 1.5544,
+      "step": 4154
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 1.5154866935152003e-10,
+      "loss": 1.5798,
+      "step": 4155
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 9.699119248196553e-11,
+      "loss": 1.5954,
+      "step": 4156
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 5.455756506345111e-11,
+      "loss": 1.3487,
+      "step": 4157
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 2.424781281984423e-11,
+      "loss": 1.4623,
+      "step": 4158
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 6.061954123115499e-12,
+      "loss": 1.6373,
+      "step": 4159
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 0.0,
+      "loss": 1.2549,
+      "step": 4160
+    },
+    {
+      "epoch": 39.71,
+      "step": 4160,
+      "total_flos": 8.527948732555076e+21,
+      "train_loss": 1.8305919360512724,
+      "train_runtime": 14627.9359,
+      "train_samples_per_second": 4.583,
+      "train_steps_per_second": 0.284
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4160,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 40,
+  "save_steps": 400,
+  "total_flos": 8.527948732555076e+21,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}