diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20808 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999134448932487,
+  "eval_steps": 500,
+  "global_step": 3463,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 1.5924,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 1.6467,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 0.4334,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.568,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 1.6116,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 1.562,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 1.5898,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.5484,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 0.4191,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.5509,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 1.4909,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 0.4192,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.4608,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.4875,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 1.3933,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.3864,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 1.3866,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.3434,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 1.3935,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.3542,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 1.3478,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.3214,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 1.2702,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.2627,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 1.2874,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.298,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 1.2878,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.2806,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 1.2555,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.2372,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 1.2217,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.2636,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 1.2652,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.225,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 1.2395,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.2536,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 1.2445,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.2443,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.2626,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.2187,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 1.2412,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.2511,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 1.2368,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.2146,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 1.2079,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.2326,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 1.2341,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.2664,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 1.2384,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 1.2012,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 1.2118,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.2208,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0192307692307692e-05,
+      "loss": 1.2868,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 1.2149,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0576923076923078e-05,
+      "loss": 1.1646,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.2002,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0961538461538464e-05,
+      "loss": 1.2006,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 1.2426,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1346153846153847e-05,
+      "loss": 1.2503,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 1.2069,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1730769230769232e-05,
+      "loss": 1.2167,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 1.239,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 1.2213,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.1916,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 1.2185,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 1.1888,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2884615384615386e-05,
+      "loss": 1.2102,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.151,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 1.1922,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 1.2222,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3653846153846156e-05,
+      "loss": 1.1905,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.1346,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.403846153846154e-05,
+      "loss": 1.1807,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 1.2312,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 1.1721,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 1.2049,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.480769230769231e-05,
+      "loss": 1.2095,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.1836,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5192307692307693e-05,
+      "loss": 0.5627,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.1868,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.557692307692308e-05,
+      "loss": 1.2017,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 1.2437,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5961538461538465e-05,
+      "loss": 1.2411,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.2034,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6346153846153847e-05,
+      "loss": 1.217,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 1.2099,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673076923076923e-05,
+      "loss": 1.1702,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.221,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7115384615384617e-05,
+      "loss": 1.1893,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 1.1662,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.1799,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.2642,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7884615384615387e-05,
+      "loss": 1.2107,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 1.1631,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.826923076923077e-05,
+      "loss": 1.2194,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.2658,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8653846153846157e-05,
+      "loss": 1.2143,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 1.1413,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.903846153846154e-05,
+      "loss": 1.1633,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.1837,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9423076923076924e-05,
+      "loss": 1.1686,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 1.1797,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.980769230769231e-05,
+      "loss": 1.2272,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.197,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999995626292754e-05,
+      "loss": 1.1893,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999982505174836e-05,
+      "loss": 1.1859,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999960636657724e-05,
+      "loss": 1.1718,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993002076055e-05,
+      "loss": 0.8747,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999890657510095e-05,
+      "loss": 1.1514,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999842546940788e-05,
+      "loss": 1.1776,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999785689094716e-05,
+      "loss": 1.2077,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999720084021615e-05,
+      "loss": 1.1934,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999645731778875e-05,
+      "loss": 1.1888,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956263243153e-05,
+      "loss": 1.1833,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999470786052272e-05,
+      "loss": 1.2244,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999937019272144e-05,
+      "loss": 1.1788,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999260852527037e-05,
+      "loss": 1.227,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999142765564693e-05,
+      "loss": 1.1548,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999901593193772e-05,
+      "loss": 1.1831,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999888035175705e-05,
+      "loss": 1.1744,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999873602514129e-05,
+      "loss": 1.1993,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999858295221668e-05,
+      "loss": 1.1682,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998421133117132e-05,
+      "loss": 1.2271,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998250567984185e-05,
+      "loss": 1.1259,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998071256967043e-05,
+      "loss": 1.2091,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997883200222557e-05,
+      "loss": 1.1777,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997686397915226e-05,
+      "loss": 1.1978,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99974808502172e-05,
+      "loss": 1.2164,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999726655730829e-05,
+      "loss": 1.1937,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997043519375934e-05,
+      "loss": 1.1607,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996811736615238e-05,
+      "loss": 1.15,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996571209228954e-05,
+      "loss": 1.207,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996321937427478e-05,
+      "loss": 1.1497,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999606392142886e-05,
+      "loss": 1.1707,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99957971614588e-05,
+      "loss": 1.1943,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995521657750638e-05,
+      "loss": 1.1944,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995237410545373e-05,
+      "loss": 1.1746,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994944420091643e-05,
+      "loss": 1.169,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994642686645746e-05,
+      "loss": 1.1658,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994332210471614e-05,
+      "loss": 0.6501,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999401299184084e-05,
+      "loss": 1.1832,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999368503103265e-05,
+      "loss": 0.6059,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993348328333933e-05,
+      "loss": 1.1863,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993002884039212e-05,
+      "loss": 1.1527,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999264869845066e-05,
+      "loss": 1.1649,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992285771878103e-05,
+      "loss": 0.5682,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991914104639004e-05,
+      "loss": 1.2125,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991533697058476e-05,
+      "loss": 1.2416,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999114454946928e-05,
+      "loss": 1.1266,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990746662211816e-05,
+      "loss": 1.1938,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990340035634134e-05,
+      "loss": 1.2557,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989924670091927e-05,
+      "loss": 1.1756,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989500565948535e-05,
+      "loss": 1.1718,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989067723574935e-05,
+      "loss": 1.2347,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988626143349753e-05,
+      "loss": 1.223,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998817582565926e-05,
+      "loss": 1.1197,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998771677089737e-05,
+      "loss": 1.1567,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998724897946563e-05,
+      "loss": 1.1666,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998677245177324e-05,
+      "loss": 1.1938,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998628718823704e-05,
+      "loss": 1.1467,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985793189281505e-05,
+      "loss": 1.1808,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985290455338765e-05,
+      "loss": 1.1703,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984778986848575e-05,
+      "loss": 1.1342,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984258784258342e-05,
+      "loss": 1.1632,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983729848023104e-05,
+      "loss": 1.1823,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998319217860555e-05,
+      "loss": 1.1721,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982645776475994e-05,
+      "loss": 1.188,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982090642112403e-05,
+      "loss": 1.1328,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981526776000374e-05,
+      "loss": 1.1591,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998095417863314e-05,
+      "loss": 1.1767,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980372850511586e-05,
+      "loss": 1.1746,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979782792144215e-05,
+      "loss": 1.1589,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979184004047174e-05,
+      "loss": 1.1834,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997857648674425e-05,
+      "loss": 1.1336,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997796024076687e-05,
+      "loss": 1.1547,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977335266654085e-05,
+      "loss": 1.123,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976701564952586e-05,
+      "loss": 1.1471,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976059136216692e-05,
+      "loss": 1.1816,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975407981008373e-05,
+      "loss": 1.1948,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974748099897218e-05,
+      "loss": 0.565,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974079493460445e-05,
+      "loss": 1.2281,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997340216228292e-05,
+      "loss": 1.1979,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997271610695713e-05,
+      "loss": 1.22,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972021328083198e-05,
+      "loss": 1.1714,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997131782626887e-05,
+      "loss": 1.1418,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970605602129537e-05,
+      "loss": 1.1575,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99698846562882e-05,
+      "loss": 0.6257,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969154989375512e-05,
+      "loss": 1.1791,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968416602029735e-05,
+      "loss": 1.1777,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996766949489677e-05,
+      "loss": 1.2218,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966913668630142e-05,
+      "loss": 1.1803,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966149123891002e-05,
+      "loss": 1.172,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965375861348135e-05,
+      "loss": 1.1356,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964593881677933e-05,
+      "loss": 1.1488,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996380318556444e-05,
+      "loss": 1.19,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963003773699305e-05,
+      "loss": 1.2118,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962195646781802e-05,
+      "loss": 1.1755,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961378805518844e-05,
+      "loss": 1.1669,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960553250624945e-05,
+      "loss": 1.1415,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995971898282226e-05,
+      "loss": 1.1849,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958876002840553e-05,
+      "loss": 1.1376,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958024311417216e-05,
+      "loss": 1.1478,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957163909297253e-05,
+      "loss": 1.1663,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956294797233303e-05,
+      "loss": 1.1818,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955416975985605e-05,
+      "loss": 1.1662,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954530446322035e-05,
+      "loss": 1.1439,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995363520901807e-05,
+      "loss": 1.1402,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952731264856812e-05,
+      "loss": 1.1619,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995181861462898e-05,
+      "loss": 1.1499,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995089725913291e-05,
+      "loss": 1.1876,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949967199174543e-05,
+      "loss": 1.185,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949028435567447e-05,
+      "loss": 1.1992,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948080969132792e-05,
+      "loss": 1.1727,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947124800699374e-05,
+      "loss": 1.1803,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946159931103586e-05,
+      "loss": 1.1446,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994518636118944e-05,
+      "loss": 1.1768,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994420409180856e-05,
+      "loss": 1.147,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943213123820186e-05,
+      "loss": 1.1228,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994221345809114e-05,
+      "loss": 1.1586,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941205095495885e-05,
+      "loss": 1.1787,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940188036916477e-05,
+      "loss": 1.2025,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993916228324257e-05,
+      "loss": 1.1923,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938127835371443e-05,
+      "loss": 1.1311,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993708469420797e-05,
+      "loss": 1.1965,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936032860664622e-05,
+      "loss": 1.1204,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993497233566149e-05,
+      "loss": 1.1253,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993390312012625e-05,
+      "loss": 1.1136,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99328252149942e-05,
+      "loss": 1.1448,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993173862120822e-05,
+      "loss": 1.1696,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99306433397188e-05,
+      "loss": 1.127,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929539371484032e-05,
+      "loss": 1.1874,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99284267174696e-05,
+      "loss": 1.1438,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927305378648788e-05,
+      "loss": 1.135,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992617535600248e-05,
+      "loss": 1.1368,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992503665051915e-05,
+      "loss": 1.1828,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923889263194876e-05,
+      "loss": 1.1849,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992273319503332e-05,
+      "loss": 1.1661,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921568447045746e-05,
+      "loss": 1.1664,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920395020251006e-05,
+      "loss": 1.1594,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919212915675542e-05,
+      "loss": 1.1585,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918022134353396e-05,
+      "loss": 1.1476,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916822677326192e-05,
+      "loss": 1.1266,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915614545643142e-05,
+      "loss": 1.1946,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914397740361052e-05,
+      "loss": 1.1189,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991317226254431e-05,
+      "loss": 1.1549,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911938113264892e-05,
+      "loss": 0.6621,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910695293602364e-05,
+      "loss": 1.1727,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990944380464386e-05,
+      "loss": 1.1672,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908183647484125e-05,
+      "loss": 1.1445,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906914823225462e-05,
+      "loss": 1.1759,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905637332977765e-05,
+      "loss": 1.131,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904351177858506e-05,
+      "loss": 1.191,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903056358992745e-05,
+      "loss": 1.1511,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901752877513108e-05,
+      "loss": 1.1702,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900440734559802e-05,
+      "loss": 1.1074,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989911993128062e-05,
+      "loss": 1.1503,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897790468830914e-05,
+      "loss": 1.151,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989645234837363e-05,
+      "loss": 1.1623,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989510557107927e-05,
+      "loss": 1.1071,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989375013812592e-05,
+      "loss": 1.1703,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892386050699236e-05,
+      "loss": 1.1701,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989101330999243e-05,
+      "loss": 1.1907,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988963191720631e-05,
+      "loss": 1.2023,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888241873549228e-05,
+      "loss": 1.1695,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886843180237117e-05,
+      "loss": 1.1175,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885435838493466e-05,
+      "loss": 1.1704,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884019849549347e-05,
+      "loss": 1.1569,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988259521464337e-05,
+      "loss": 1.1499,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881161935021732e-05,
+      "loss": 1.1617,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987972001193818e-05,
+      "loss": 0.6458,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878269446654023e-05,
+      "loss": 0.554,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876810240438132e-05,
+      "loss": 1.2378,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987534239456693e-05,
+      "loss": 1.2089,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987386591032441e-05,
+      "loss": 1.1807,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872380789002105e-05,
+      "loss": 1.2163,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870887031899123e-05,
+      "loss": 1.1715,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869384640322108e-05,
+      "loss": 1.1623,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867873615585266e-05,
+      "loss": 1.1395,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866353959010355e-05,
+      "loss": 1.142,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864825671926675e-05,
+      "loss": 1.1883,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986328875567109e-05,
+      "loss": 1.1621,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861743211588e-05,
+      "loss": 1.1684,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860189041029358e-05,
+      "loss": 1.1154,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985862624535466e-05,
+      "loss": 1.17,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857054825930952e-05,
+      "loss": 1.1476,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855474784132815e-05,
+      "loss": 1.1545,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853886121342377e-05,
+      "loss": 1.1335,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985228883894931e-05,
+      "loss": 1.1479,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850682938350824e-05,
+      "loss": 1.1743,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849068420951663e-05,
+      "loss": 1.1734,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847445288164114e-05,
+      "loss": 1.159,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845813541407998e-05,
+      "loss": 1.1827,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844173182110676e-05,
+      "loss": 1.1863,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984252421170703e-05,
+      "loss": 1.1554,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840866631639488e-05,
+      "loss": 1.1243,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839200443358005e-05,
+      "loss": 1.1565,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837525648320064e-05,
+      "loss": 1.1573,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835842247990677e-05,
+      "loss": 1.131,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983415024384238e-05,
+      "loss": 1.1551,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832449637355248e-05,
+      "loss": 1.1441,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830740430016864e-05,
+      "loss": 1.155,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829022623322343e-05,
+      "loss": 1.144,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982729621877433e-05,
+      "loss": 1.2151,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825561217882975e-05,
+      "loss": 1.1485,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982381762216596e-05,
+      "loss": 1.114,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982206543314847e-05,
+      "loss": 1.1737,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982030465236323e-05,
+      "loss": 1.1589,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818535281350463e-05,
+      "loss": 0.7058,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816757321657908e-05,
+      "loss": 0.5018,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814970774840828e-05,
+      "loss": 1.2046,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813175642461982e-05,
+      "loss": 1.136,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811371926091647e-05,
+      "loss": 1.1956,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809559627307612e-05,
+      "loss": 1.1636,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980773874769517e-05,
+      "loss": 1.1934,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980590928884712e-05,
+      "loss": 0.7511,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804071252363763e-05,
+      "loss": 1.207,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802224639852904e-05,
+      "loss": 1.1928,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800369452929855e-05,
+      "loss": 1.2193,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798505693217426e-05,
+      "loss": 1.2167,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979663336234592e-05,
+      "loss": 1.2008,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979475246195315e-05,
+      "loss": 1.2466,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792862993684407e-05,
+      "loss": 1.1702,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790964959192495e-05,
+      "loss": 1.1985,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789058360137706e-05,
+      "loss": 1.1472,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787143198187814e-05,
+      "loss": 1.1793,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785219475018093e-05,
+      "loss": 1.171,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97832871923113e-05,
+      "loss": 1.1394,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781346351757695e-05,
+      "loss": 1.1929,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779396955054996e-05,
+      "loss": 1.1624,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777439003908432e-05,
+      "loss": 1.1413,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97754725000307e-05,
+      "loss": 1.1972,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977349744514198e-05,
+      "loss": 1.1329,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771513840969937e-05,
+      "loss": 1.1647,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769521689249715e-05,
+      "loss": 0.6071,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767520991723927e-05,
+      "loss": 1.1296,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976551175014267e-05,
+      "loss": 1.14,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763493966263506e-05,
+      "loss": 1.2054,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761467641851478e-05,
+      "loss": 1.1533,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759432778679093e-05,
+      "loss": 1.1738,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757389378526334e-05,
+      "loss": 1.1308,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755337443180643e-05,
+      "loss": 1.1675,
+      "step": 341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975327697443694e-05,
+      "loss": 1.145,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751207974097594e-05,
+      "loss": 1.1689,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974913044397245e-05,
+      "loss": 1.1865,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747044385878808e-05,
+      "loss": 1.1618,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744949801641433e-05,
+      "loss": 1.1511,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742846693092542e-05,
+      "loss": 1.1294,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740735062071812e-05,
+      "loss": 1.124,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738614910426373e-05,
+      "loss": 1.1614,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973648624001081e-05,
+      "loss": 1.1308,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973434905268716e-05,
+      "loss": 1.0987,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732203350324904e-05,
+      "loss": 1.1245,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730049134800987e-05,
+      "loss": 1.1534,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727886407999782e-05,
+      "loss": 1.1857,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972571517181312e-05,
+      "loss": 1.1487,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972353542814027e-05,
+      "loss": 1.1931,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721347178887947e-05,
+      "loss": 1.1294,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719150425970297e-05,
+      "loss": 1.1776,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716945171308918e-05,
+      "loss": 1.2021,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714731416832832e-05,
+      "loss": 1.1296,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97125091644785e-05,
+      "loss": 1.1726,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710278416189828e-05,
+      "loss": 1.1362,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708039173918136e-05,
+      "loss": 1.1188,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705791439622187e-05,
+      "loss": 1.1629,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970353521526816e-05,
+      "loss": 1.1609,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701270502829676e-05,
+      "loss": 1.1395,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698997304287764e-05,
+      "loss": 1.1231,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696715621630895e-05,
+      "loss": 1.1443,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694425456854946e-05,
+      "loss": 1.159,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969212681196322e-05,
+      "loss": 1.1351,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968981968896644e-05,
+      "loss": 1.1266,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687504089882734e-05,
+      "loss": 1.1636,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685180016737657e-05,
+      "loss": 1.1152,
+      "step": 373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682847471564176e-05,
+      "loss": 1.1271,
+      "step": 374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968050645640266e-05,
+      "loss": 1.1399,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967815697330089e-05,
+      "loss": 1.1532,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675799024314063e-05,
+      "loss": 1.1933,
+      "step": 377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673432611504772e-05,
+      "loss": 1.1156,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671057736943016e-05,
+      "loss": 1.1511,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668674402706195e-05,
+      "loss": 1.1431,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966628261087911e-05,
+      "loss": 1.1373,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966388236355396e-05,
+      "loss": 0.4819,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966147366283035e-05,
+      "loss": 1.1649,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659056510815256e-05,
+      "loss": 1.1272,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965663090962307e-05,
+      "loss": 1.0912,
+      "step": 385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654196861375564e-05,
+      "loss": 1.09,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651754368201903e-05,
+      "loss": 1.1268,
+      "step": 387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649303432238633e-05,
+      "loss": 1.1373,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964684405562969e-05,
+      "loss": 1.1838,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644376240526392e-05,
+      "loss": 1.1444,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641899989087447e-05,
+      "loss": 1.112,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639415303478923e-05,
+      "loss": 1.1376,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636922185874284e-05,
+      "loss": 1.1447,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634420638454365e-05,
+      "loss": 0.5142,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631910663407366e-05,
+      "loss": 1.1635,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629392262928873e-05,
+      "loss": 1.1463,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626865439221838e-05,
+      "loss": 1.1689,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624330194496566e-05,
+      "loss": 1.1121,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962178653097075e-05,
+      "loss": 1.1078,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961923445086944e-05,
+      "loss": 1.1326,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616673956425037e-05,
+      "loss": 1.1209,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961410504987732e-05,
+      "loss": 1.1432,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611527733473416e-05,
+      "loss": 1.1165,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960894200946781e-05,
+      "loss": 1.1742,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960634788012234e-05,
+      "loss": 1.0954,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603745347706198e-05,
+      "loss": 1.1209,
+      "step": 406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601134414495933e-05,
+      "loss": 1.1527,
+      "step": 407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598515082775427e-05,
+      "loss": 1.1455,
+      "step": 408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959588735483593e-05,
+      "loss": 1.0662,
+      "step": 409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593251232976013e-05,
+      "loss": 1.1533,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959060671950161e-05,
+      "loss": 0.4787,
+      "step": 411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958795381672598e-05,
+      "loss": 1.1287,
+      "step": 412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958529252696973e-05,
+      "loss": 1.1137,
+      "step": 413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95826228525608e-05,
+      "loss": 1.1512,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579944795834463e-05,
+      "loss": 1.1786,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577258359133335e-05,
+      "loss": 0.4271,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957456354480734e-05,
+      "loss": 1.1639,
+      "step": 417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957186035521375e-05,
+      "loss": 1.1657,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956914879271716e-05,
+      "loss": 1.1303,
+      "step": 419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566428859689486e-05,
+      "loss": 1.1272,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956370055850996e-05,
+      "loss": 1.1879,
+      "step": 421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560963891565145e-05,
+      "loss": 1.1134,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558218861248916e-05,
+      "loss": 1.1548,
+      "step": 423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555465469962465e-05,
+      "loss": 1.1652,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552703720114298e-05,
+      "loss": 1.1138,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954993361412023e-05,
+      "loss": 1.1189,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547155154403388e-05,
+      "loss": 1.1604,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544368343394205e-05,
+      "loss": 0.4446,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541573183530426e-05,
+      "loss": 1.1308,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538769677257087e-05,
+      "loss": 1.1197,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535957827026534e-05,
+      "loss": 1.1569,
+      "step": 431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533137635298406e-05,
+      "loss": 1.1539,
+      "step": 432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9530309104539647e-05,
+      "loss": 1.0855,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952747223722448e-05,
+      "loss": 1.0654,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524627035834447e-05,
+      "loss": 1.1162,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521773502858347e-05,
+      "loss": 1.1191,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518911640792294e-05,
+      "loss": 1.1445,
+      "step": 437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516041452139668e-05,
+      "loss": 0.4336,
+      "step": 438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513162939411154e-05,
+      "loss": 1.161,
+      "step": 439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510276105124696e-05,
+      "loss": 1.1371,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507380951805533e-05,
+      "loss": 1.1683,
+      "step": 441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504477481986175e-05,
+      "loss": 1.1675,
+      "step": 442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501565698206406e-05,
+      "loss": 1.1293,
+      "step": 443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498645603013285e-05,
+      "loss": 1.1514,
+      "step": 444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495717198961138e-05,
+      "loss": 1.145,
+      "step": 445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492780488611568e-05,
+      "loss": 0.4189,
+      "step": 446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489835474533427e-05,
+      "loss": 1.1588,
+      "step": 447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948688215930285e-05,
+      "loss": 1.1053,
+      "step": 448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483920545503217e-05,
+      "loss": 1.1152,
+      "step": 449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480950635725185e-05,
+      "loss": 1.111,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477972432566644e-05,
+      "loss": 1.0903,
+      "step": 451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947498593863276e-05,
+      "loss": 1.1586,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471991156535942e-05,
+      "loss": 1.0857,
+      "step": 453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468988088895847e-05,
+      "loss": 1.1848,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465976738339386e-05,
+      "loss": 1.1438,
+      "step": 455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462957107500714e-05,
+      "loss": 1.1283,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459929199021222e-05,
+      "loss": 1.119,
+      "step": 457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456893015549547e-05,
+      "loss": 1.1467,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945384855974157e-05,
+      "loss": 1.1023,
+      "step": 459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450795834260395e-05,
+      "loss": 1.1391,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944773484177638e-05,
+      "loss": 1.097,
+      "step": 461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444665584967087e-05,
+      "loss": 1.1016,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944158806651733e-05,
+      "loss": 1.091,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438502289119144e-05,
+      "loss": 1.1586,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435408255471783e-05,
+      "loss": 1.1025,
+      "step": 465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432305968281723e-05,
+      "loss": 1.0812,
+      "step": 466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942919543026267e-05,
+      "loss": 0.4365,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942607664413554e-05,
+      "loss": 1.0945,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942294961262846e-05,
+      "loss": 1.1199,
+      "step": 469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419814338476776e-05,
+      "loss": 1.143,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416670824423044e-05,
+      "loss": 1.1513,
+      "step": 471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413519073217025e-05,
+      "loss": 1.1521,
+      "step": 472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410359087615683e-05,
+      "loss": 1.1425,
+      "step": 473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940719087038319e-05,
+      "loss": 0.4029,
+      "step": 474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404014424290923e-05,
+      "loss": 1.1592,
+      "step": 475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400829752117446e-05,
+      "loss": 1.1843,
+      "step": 476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397636856648523e-05,
+      "loss": 1.1475,
+      "step": 477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394435740677113e-05,
+      "loss": 1.1541,
+      "step": 478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391226407003366e-05,
+      "loss": 1.1374,
+      "step": 479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388008858434617e-05,
+      "loss": 1.163,
+      "step": 480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938478309778539e-05,
+      "loss": 1.1437,
+      "step": 481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381549127877392e-05,
+      "loss": 1.1437,
+      "step": 482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378306951539513e-05,
+      "loss": 1.1024,
+      "step": 483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375056571607814e-05,
+      "loss": 1.0738,
+      "step": 484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937179799092554e-05,
+      "loss": 1.1778,
+      "step": 485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368531212343104e-05,
+      "loss": 0.4262,
+      "step": 486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365256238718098e-05,
+      "loss": 1.1791,
+      "step": 487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936197307291527e-05,
+      "loss": 1.1527,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358681717806548e-05,
+      "loss": 0.3809,
+      "step": 489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355382176271014e-05,
+      "loss": 1.181,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352074451194914e-05,
+      "loss": 1.1408,
+      "step": 491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934875854547165e-05,
+      "loss": 1.0947,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345434462001784e-05,
+      "loss": 1.1017,
+      "step": 493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342102203693027e-05,
+      "loss": 1.1775,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933876177346025e-05,
+      "loss": 1.1016,
+      "step": 495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933541317422546e-05,
+      "loss": 1.1455,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332056408917817e-05,
+      "loss": 1.1293,
+      "step": 497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328691480473627e-05,
+      "loss": 1.0948,
+      "step": 498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325318391836325e-05,
+      "loss": 1.0992,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321937145956496e-05,
+      "loss": 1.1685,
+      "step": 500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318547745791853e-05,
+      "loss": 1.0993,
+      "step": 501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315150194307248e-05,
+      "loss": 1.138,
+      "step": 502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931174449447466e-05,
+      "loss": 1.1364,
+      "step": 503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308330649273192e-05,
+      "loss": 1.1688,
+      "step": 504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930490866168908e-05,
+      "loss": 1.1196,
+      "step": 505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301478534715675e-05,
+      "loss": 1.0996,
+      "step": 506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298040271353456e-05,
+      "loss": 1.1074,
+      "step": 507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929459387461001e-05,
+      "loss": 1.1562,
+      "step": 508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291139347500046e-05,
+      "loss": 1.15,
+      "step": 509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928767669304538e-05,
+      "loss": 1.1126,
+      "step": 510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284205914274937e-05,
+      "loss": 1.1756,
+      "step": 511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280727014224757e-05,
+      "loss": 1.1476,
+      "step": 512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277239995937974e-05,
+      "loss": 1.145,
+      "step": 513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273744862464832e-05,
+      "loss": 1.0764,
+      "step": 514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270241616862663e-05,
+      "loss": 1.1182,
+      "step": 515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266730262195903e-05,
+      "loss": 1.1475,
+      "step": 516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926321080153608e-05,
+      "loss": 1.1128,
+      "step": 517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259683237961816e-05,
+      "loss": 1.1253,
+      "step": 518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925614757455881e-05,
+      "loss": 1.1697,
+      "step": 519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925260381441986e-05,
+      "loss": 1.0977,
+      "step": 520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249051960644837e-05,
+      "loss": 1.0822,
+      "step": 521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924549201634069e-05,
+      "loss": 0.4065,
+      "step": 522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241923984621458e-05,
+      "loss": 1.1012,
+      "step": 523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238347868608242e-05,
+      "loss": 1.131,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923476367142922e-05,
+      "loss": 1.1485,
+      "step": 525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231171396219635e-05,
+      "loss": 1.1237,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227571046121804e-05,
+      "loss": 1.0897,
+      "step": 527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92239626242851e-05,
+      "loss": 1.0824,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220346133865953e-05,
+      "loss": 1.1364,
+      "step": 529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921672157802787e-05,
+      "loss": 1.0909,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921308895994139e-05,
+      "loss": 1.1063,
+      "step": 531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920944828278412e-05,
+      "loss": 0.4016,
+      "step": 532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205799549740706e-05,
+      "loss": 0.3914,
+      "step": 533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202142764002853e-05,
+      "loss": 1.0985,
+      "step": 534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198477928769295e-05,
+      "loss": 0.4017,
+      "step": 535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194805047245822e-05,
+      "loss": 1.1395,
+      "step": 536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9191124122645254e-05,
+      "loss": 1.1269,
+      "step": 537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187435158187443e-05,
+      "loss": 1.1261,
+      "step": 538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918373815709929e-05,
+      "loss": 1.079,
+      "step": 539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180033122614703e-05,
+      "loss": 1.1029,
+      "step": 540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176320057974638e-05,
+      "loss": 1.1417,
+      "step": 541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172598966427064e-05,
+      "loss": 1.1142,
+      "step": 542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916886985122697e-05,
+      "loss": 1.087,
+      "step": 543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165132715636373e-05,
+      "loss": 1.1048,
+      "step": 544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161387562924297e-05,
+      "loss": 0.418,
+      "step": 545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157634396366788e-05,
+      "loss": 1.0916,
+      "step": 546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915387321924689e-05,
+      "loss": 1.0996,
+      "step": 547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150104034854663e-05,
+      "loss": 1.1107,
+      "step": 548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146326846487168e-05,
+      "loss": 1.1222,
+      "step": 549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142541657448468e-05,
+      "loss": 1.1183,
+      "step": 550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138748471049626e-05,
+      "loss": 0.4004,
+      "step": 551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91349472906087e-05,
+      "loss": 1.1302,
+      "step": 552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131138119450735e-05,
+      "loss": 0.3976,
+      "step": 553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912732096090778e-05,
+      "loss": 1.059,
+      "step": 554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123495818318856e-05,
+      "loss": 1.1671,
+      "step": 555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119662695029973e-05,
+      "loss": 1.0762,
+      "step": 556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115821594394124e-05,
+      "loss": 1.1669,
+      "step": 557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111972519771278e-05,
+      "loss": 1.0997,
+      "step": 558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108115474528384e-05,
+      "loss": 1.1653,
+      "step": 559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104250462039355e-05,
+      "loss": 1.1228,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100377485685078e-05,
+      "loss": 1.1488,
+      "step": 561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909649654885341e-05,
+      "loss": 1.13,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092607654939166e-05,
+      "loss": 1.1172,
+      "step": 563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088710807344117e-05,
+      "loss": 0.4137,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084806009477005e-05,
+      "loss": 1.1442,
+      "step": 565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080893264753514e-05,
+      "loss": 1.1461,
+      "step": 566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076972576596283e-05,
+      "loss": 1.1019,
+      "step": 567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073043948434903e-05,
+      "loss": 1.0955,
+      "step": 568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069107383705907e-05,
+      "loss": 1.116,
+      "step": 569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906516288585277e-05,
+      "loss": 1.1329,
+      "step": 570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061210458325913e-05,
+      "loss": 1.1259,
+      "step": 571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905725010458268e-05,
+      "loss": 1.1421,
+      "step": 572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053281828087365e-05,
+      "loss": 1.1121,
+      "step": 573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904930563231118e-05,
+      "loss": 1.1606,
+      "step": 574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045321520732267e-05,
+      "loss": 1.1036,
+      "step": 575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041329496835696e-05,
+      "loss": 1.113,
+      "step": 576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037329564113454e-05,
+      "loss": 1.1014,
+      "step": 577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903332172606445e-05,
+      "loss": 1.0717,
+      "step": 578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029305986194502e-05,
+      "loss": 1.0637,
+      "step": 579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025282348016347e-05,
+      "loss": 1.1606,
+      "step": 580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021250815049627e-05,
+      "loss": 1.1023,
+      "step": 581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901721139082089e-05,
+      "loss": 1.0984,
+      "step": 582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013164078863597e-05,
+      "loss": 1.1458,
+      "step": 583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900910888271809e-05,
+      "loss": 1.1277,
+      "step": 584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005045805931614e-05,
+      "loss": 1.1273,
+      "step": 585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900097485205832e-05,
+      "loss": 1.1135,
+      "step": 586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996896024659237e-05,
+      "loss": 1.1084,
+      "step": 587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992809327302282e-05,
+      "loss": 1.1444,
+      "step": 588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898871476356226e-05,
+      "loss": 1.08,
+      "step": 589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984612337020854e-05,
+      "loss": 1.1185,
+      "step": 590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898050205126663e-05,
+      "loss": 1.0779,
+      "step": 591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976383909895023e-05,
+      "loss": 1.192,
+      "step": 592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897225791650834e-05,
+      "loss": 1.1893,
+      "step": 593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968124074715768e-05,
+      "loss": 1.1457,
+      "step": 594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963982388133337e-05,
+      "loss": 1.1183,
+      "step": 595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895983286038396e-05,
+      "loss": 1.1125,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955675495097395e-05,
+      "loss": 1.1554,
+      "step": 597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895151029591027e-05,
+      "loss": 1.1299,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947337266466047e-05,
+      "loss": 1.1532,
+      "step": 599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943156410415057e-05,
+      "loss": 1.1088,
+      "step": 600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938967731414467e-05,
+      "loss": 1.1121,
+      "step": 601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893477123312828e-05,
+      "loss": 1.1489,
+      "step": 602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893056691922736e-05,
+      "loss": 1.0918,
+      "step": 603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926354793389378e-05,
+      "loss": 1.1059,
+      "step": 604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892213485929887e-05,
+      "loss": 1.1214,
+      "step": 605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891790712064718e-05,
+      "loss": 1.1002,
+      "step": 606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913671581132492e-05,
+      "loss": 1.126,
+      "step": 607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89094282444598e-05,
+      "loss": 1.1202,
+      "step": 608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890517711434093e-05,
+      "loss": 1.1289,
+      "step": 609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890091819449452e-05,
+      "loss": 1.0881,
+      "step": 610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889665148864603e-05,
+      "loss": 0.3918,
+      "step": 611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889237700052772e-05,
+      "loss": 1.1709,
+      "step": 612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888809473387866e-05,
+      "loss": 1.106,
+      "step": 613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888380469244473e-05,
+      "loss": 1.0855,
+      "step": 614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879506879978603e-05,
+      "loss": 1.1544,
+      "step": 615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875201300239757e-05,
+      "loss": 1.1156,
+      "step": 616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887088795699446e-05,
+      "loss": 0.3881,
+      "step": 617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866566854015772e-05,
+      "loss": 1.0941,
+      "step": 618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862237995083544e-05,
+      "loss": 1.1104,
+      "step": 619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857901383984403e-05,
+      "loss": 1.1638,
+      "step": 620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853557024511766e-05,
+      "loss": 1.1315,
+      "step": 621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884920492046582e-05,
+      "loss": 1.1161,
+      "step": 622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884484507565354e-05,
+      "loss": 1.1624,
+      "step": 623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840477493888653e-05,
+      "loss": 1.1374,
+      "step": 624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836102178991668e-05,
+      "loss": 1.1172,
+      "step": 625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831719134789854e-05,
+      "loss": 1.093,
+      "step": 626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827328365117243e-05,
+      "loss": 1.1019,
+      "step": 627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882292987381462e-05,
+      "loss": 1.0931,
+      "step": 628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818523664729528e-05,
+      "loss": 1.0782,
+      "step": 629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814109741716263e-05,
+      "loss": 1.1054,
+      "step": 630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880968810863587e-05,
+      "loss": 1.1358,
+      "step": 631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805258769356126e-05,
+      "loss": 1.1478,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800821727751564e-05,
+      "loss": 1.0662,
+      "step": 633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796376987703445e-05,
+      "loss": 1.1277,
+      "step": 634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879192455309977e-05,
+      "loss": 1.0682,
+      "step": 635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787464427835265e-05,
+      "loss": 1.1017,
+      "step": 636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878299661581139e-05,
+      "loss": 1.1388,
+      "step": 637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778521120936322e-05,
+      "loss": 1.1486,
+      "step": 638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774037947124962e-05,
+      "loss": 1.0606,
+      "step": 639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876954709829893e-05,
+      "loss": 1.1427,
+      "step": 640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8765048578386557e-05,
+      "loss": 1.1445,
+      "step": 641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876054239132289e-05,
+      "loss": 0.4007,
+      "step": 642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756028541049662e-05,
+      "loss": 1.0975,
+      "step": 643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751507031515342e-05,
+      "loss": 1.1483,
+      "step": 644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746977866675074e-05,
+      "loss": 1.0899,
+      "step": 645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742441050490706e-05,
+      "loss": 0.3838,
+      "step": 646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873789658693078e-05,
+      "loss": 1.0993,
+      "step": 647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873334447997053e-05,
+      "loss": 1.133,
+      "step": 648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728784733591867e-05,
+      "loss": 1.1324,
+      "step": 649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724217351783394e-05,
+      "loss": 1.0374,
+      "step": 650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719642338540388e-05,
+      "loss": 1.0495,
+      "step": 651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715059697864805e-05,
+      "loss": 1.1269,
+      "step": 652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710469433765266e-05,
+      "loss": 1.1621,
+      "step": 653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705871550257065e-05,
+      "loss": 1.107,
+      "step": 654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701266051362167e-05,
+      "loss": 1.1142,
+      "step": 655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696652941109192e-05,
+      "loss": 1.0672,
+      "step": 656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692032223533415e-05,
+      "loss": 1.0879,
+      "step": 657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868740390267677e-05,
+      "loss": 1.086,
+      "step": 658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682767982587844e-05,
+      "loss": 1.1143,
+      "step": 659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678124467321862e-05,
+      "loss": 1.1416,
+      "step": 660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673473360940705e-05,
+      "loss": 1.0938,
+      "step": 661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866881466751289e-05,
+      "loss": 1.1512,
+      "step": 662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664148391113565e-05,
+      "loss": 1.1703,
+      "step": 663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659474535824517e-05,
+      "loss": 1.1019,
+      "step": 664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865479310573416e-05,
+      "loss": 1.1382,
+      "step": 665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650104104937537e-05,
+      "loss": 1.0784,
+      "step": 666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864540753753631e-05,
+      "loss": 1.1074,
+      "step": 667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640703407638757e-05,
+      "loss": 1.0038,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635991719359784e-05,
+      "loss": 1.054,
+      "step": 669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631272476820893e-05,
+      "loss": 1.1124,
+      "step": 670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626545684150207e-05,
+      "loss": 1.1287,
+      "step": 671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862181134548244e-05,
+      "loss": 1.0955,
+      "step": 672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617069464958923e-05,
+      "loss": 1.094,
+      "step": 673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612320046727568e-05,
+      "loss": 1.0945,
+      "step": 674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8607563094942893e-05,
+      "loss": 1.1318,
+      "step": 675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602798613765995e-05,
+      "loss": 1.1291,
+      "step": 676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859802660736457e-05,
+      "loss": 1.134,
+      "step": 677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593247079912885e-05,
+      "loss": 1.1012,
+      "step": 678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588460035591796e-05,
+      "loss": 1.1213,
+      "step": 679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858366547858872e-05,
+      "loss": 1.1,
+      "step": 680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578863413097665e-05,
+      "loss": 1.0964,
+      "step": 681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857405384331919e-05,
+      "loss": 1.1041,
+      "step": 682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569236773460428e-05,
+      "loss": 1.092,
+      "step": 683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856441220773507e-05,
+      "loss": 1.1049,
+      "step": 684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559580150363357e-05,
+      "loss": 1.135,
+      "step": 685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85547406055721e-05,
+      "loss": 1.068,
+      "step": 686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549893577594648e-05,
+      "loss": 1.0819,
+      "step": 687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545039070670886e-05,
+      "loss": 1.093,
+      "step": 688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540177089047265e-05,
+      "loss": 1.1214,
+      "step": 689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535307636976756e-05,
+      "loss": 1.1188,
+      "step": 690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530430718718874e-05,
+      "loss": 1.1089,
+      "step": 691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852554633853966e-05,
+      "loss": 1.0903,
+      "step": 692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520654500711676e-05,
+      "loss": 1.1145,
+      "step": 693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515755209514026e-05,
+      "loss": 1.1245,
+      "step": 694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510848469232323e-05,
+      "loss": 1.0918,
+      "step": 695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850593428415869e-05,
+      "loss": 1.1097,
+      "step": 696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501012658591773e-05,
+      "loss": 1.1003,
+      "step": 697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849608359683672e-05,
+      "loss": 1.1459,
+      "step": 698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491147103205183e-05,
+      "loss": 1.1014,
+      "step": 699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848620318201532e-05,
+      "loss": 1.1502,
+      "step": 700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481251837591784e-05,
+      "loss": 1.1317,
+      "step": 701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476293074265723e-05,
+      "loss": 1.0949,
+      "step": 702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847132689637477e-05,
+      "loss": 1.0537,
+      "step": 703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466353308263046e-05,
+      "loss": 1.1237,
+      "step": 704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846137231428116e-05,
+      "loss": 1.1386,
+      "step": 705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845638391878619e-05,
+      "loss": 1.1185,
+      "step": 706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845138812614169e-05,
+      "loss": 1.1125,
+      "step": 707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446384940717688e-05,
+      "loss": 1.143,
+      "step": 708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844137436689068e-05,
+      "loss": 1.1186,
+      "step": 709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8436356409043626e-05,
+      "loss": 1.1237,
+      "step": 710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431331071565935e-05,
+      "loss": 1.1131,
+      "step": 711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426298358853478e-05,
+      "loss": 1.1359,
+      "step": 712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842125827530858e-05,
+      "loss": 1.0982,
+      "step": 713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416210825340012e-05,
+      "loss": 1.1136,
+      "step": 714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841115601336299e-05,
+      "loss": 1.1031,
+      "step": 715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840609384379916e-05,
+      "loss": 1.066,
+      "step": 716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401024321076615e-05,
+      "loss": 1.0978,
+      "step": 717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839594744962988e-05,
+      "loss": 1.0682,
+      "step": 718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83908632338999e-05,
+      "loss": 1.1395,
+      "step": 719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838577167833405e-05,
+      "loss": 1.0916,
+      "step": 720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838067278738613e-05,
+      "loss": 1.1264,
+      "step": 721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375566565516344e-05,
+      "loss": 1.0865,
+      "step": 722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837045301719132e-05,
+      "loss": 1.0826,
+      "step": 723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836533214688409e-05,
+      "loss": 0.3982,
+      "step": 724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836020395907409e-05,
+      "loss": 1.1538,
+      "step": 725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835506845824716e-05,
+      "loss": 1.1304,
+      "step": 726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349925648895543e-05,
+      "loss": 1.1046,
+      "step": 727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834477553551785e-05,
+      "loss": 1.1212,
+      "step": 728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339618122619114e-05,
+      "loss": 1.1581,
+      "step": 729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334453414710733e-05,
+      "loss": 1.0994,
+      "step": 730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329281416310485e-05,
+      "loss": 1.075,
+      "step": 731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324102131942537e-05,
+      "loss": 1.0813,
+      "step": 732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831891556613743e-05,
+      "loss": 1.0966,
+      "step": 733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313721723432058e-05,
+      "loss": 1.0659,
+      "step": 734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308520608369692e-05,
+      "loss": 0.4409,
+      "step": 735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830331222549997e-05,
+      "loss": 1.1404,
+      "step": 736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298096579378867e-05,
+      "loss": 1.1214,
+      "step": 737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829287367456874e-05,
+      "loss": 1.0677,
+      "step": 738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828764351563827e-05,
+      "loss": 1.0647,
+      "step": 739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82824061071625e-05,
+      "loss": 1.0871,
+      "step": 740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277161453722805e-05,
+      "loss": 1.0923,
+      "step": 741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271909559906903e-05,
+      "loss": 1.1019,
+      "step": 742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266650430308837e-05,
+      "loss": 1.1578,
+      "step": 743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261384069528995e-05,
+      "loss": 1.1036,
+      "step": 744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8256110482174075e-05,
+      "loss": 1.0739,
+      "step": 745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82508296728571e-05,
+      "loss": 1.0453,
+      "step": 746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824554164619742e-05,
+      "loss": 1.1082,
+      "step": 747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240246406820686e-05,
+      "loss": 1.1235,
+      "step": 748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234943959358865e-05,
+      "loss": 1.1077,
+      "step": 749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822963430845023e-05,
+      "loss": 1.1198,
+      "step": 750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822431745873935e-05,
+      "loss": 1.1705,
+      "step": 751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821899341487709e-05,
+      "loss": 1.0946,
+      "step": 752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213662181520622e-05,
+      "loss": 1.0732,
+      "step": 753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208323763333386e-05,
+      "loss": 1.1442,
+      "step": 754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202978164985126e-05,
+      "loss": 1.1459,
+      "step": 755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197625391151855e-05,
+      "loss": 1.1481,
+      "step": 756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819226544651586e-05,
+      "loss": 1.1168,
+      "step": 757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818689833576572e-05,
+      "loss": 1.1295,
+      "step": 758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181524063596258e-05,
+      "loss": 1.1229,
+      "step": 759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176142634708576e-05,
+      "loss": 1.1427,
+      "step": 760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170754053810034e-05,
+      "loss": 1.1405,
+      "step": 761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816535832561425e-05,
+      "loss": 1.0859,
+      "step": 762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159955454841086e-05,
+      "loss": 1.1079,
+      "step": 763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815454544621666e-05,
+      "loss": 1.1232,
+      "step": 764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149128304473323e-05,
+      "loss": 1.1274,
+      "step": 765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143704034349688e-05,
+      "loss": 1.0729,
+      "step": 766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138272640590576e-05,
+      "loss": 1.0829,
+      "step": 767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132834127947057e-05,
+      "loss": 1.1318,
+      "step": 768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127388501176428e-05,
+      "loss": 1.1087,
+      "step": 769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121935765042197e-05,
+      "loss": 1.1202,
+      "step": 770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116475924314102e-05,
+      "loss": 1.0557,
+      "step": 771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811100898376809e-05,
+      "loss": 1.0876,
+      "step": 772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105534948186324e-05,
+      "loss": 1.1462,
+      "step": 773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100053822357168e-05,
+      "loss": 1.0991,
+      "step": 774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094565611075188e-05,
+      "loss": 1.0652,
+      "step": 775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089070319141148e-05,
+      "loss": 1.0795,
+      "step": 776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083567951362016e-05,
+      "loss": 1.1328,
+      "step": 777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078058512550933e-05,
+      "loss": 1.0883,
+      "step": 778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072542007527238e-05,
+      "loss": 1.1212,
+      "step": 779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067018441116447e-05,
+      "loss": 1.1312,
+      "step": 780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806148781815025e-05,
+      "loss": 1.0937,
+      "step": 781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805595014346651e-05,
+      "loss": 1.1456,
+      "step": 782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805040542190927e-05,
+      "loss": 1.07,
+      "step": 783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044853658328717e-05,
+      "loss": 1.0804,
+      "step": 784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803929485758121e-05,
+      "loss": 1.108,
+      "step": 785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803372902452927e-05,
+      "loss": 1.0411,
+      "step": 786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028156164041558e-05,
+      "loss": 1.0077,
+      "step": 787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022576280992884e-05,
+      "loss": 1.103,
+      "step": 788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016989380264204e-05,
+      "loss": 1.0891,
+      "step": 789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011395466742613e-05,
+      "loss": 1.1497,
+      "step": 790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005794545321336e-05,
+      "loss": 1.0985,
+      "step": 791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000186620899734e-05,
+      "loss": 1.1188,
+      "step": 792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994571698383293e-05,
+      "loss": 1.1023,
+      "step": 793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798894978268361e-05,
+      "loss": 1.142,
+      "step": 794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983320878718416e-05,
+      "loss": 1.1589,
+      "step": 795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977684991411545e-05,
+      "loss": 1.0531,
+      "step": 796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972042125692942e-05,
+      "loss": 0.4614,
+      "step": 797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966392286498648e-05,
+      "loss": 1.0965,
+      "step": 798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960735478770823e-05,
+      "loss": 1.0846,
+      "step": 799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955071707457704e-05,
+      "loss": 1.0834,
+      "step": 800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949400977513624e-05,
+      "loss": 1.0908,
+      "step": 801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943723293899015e-05,
+      "loss": 1.1267,
+      "step": 802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793803866158037e-05,
+      "loss": 1.103,
+      "step": 803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932347085530283e-05,
+      "loss": 1.1536,
+      "step": 804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792664857072741e-05,
+      "loss": 1.1054,
+      "step": 805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920943122156473e-05,
+      "loss": 1.0929,
+      "step": 806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915230744808267e-05,
+      "loss": 1.0638,
+      "step": 807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909511443679644e-05,
+      "loss": 1.0919,
+      "step": 808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790378522377352e-05,
+      "loss": 1.0671,
+      "step": 809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898052090098846e-05,
+      "loss": 1.1235,
+      "step": 810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892312047670643e-05,
+      "loss": 1.0698,
+      "step": 811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886565101509957e-05,
+      "loss": 0.5313,
+      "step": 812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7880811256643878e-05,
+      "loss": 1.0853,
+      "step": 813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787505051810554e-05,
+      "loss": 1.111,
+      "step": 814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869282890934095e-05,
+      "loss": 1.1225,
+      "step": 815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786350838017473e-05,
+      "loss": 1.0872,
+      "step": 816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785772699087864e-05,
+      "loss": 1.1148,
+      "step": 817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851938728103055e-05,
+      "loss": 1.1067,
+      "step": 818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846143596911205e-05,
+      "loss": 1.1264,
+      "step": 819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784034160237233e-05,
+      "loss": 1.0492,
+      "step": 820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834532749561675e-05,
+      "loss": 1.1183,
+      "step": 821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782871704356049e-05,
+      "loss": 1.1006,
+      "step": 822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822894489456005e-05,
+      "loss": 0.4765,
+      "step": 823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817065092341455e-05,
+      "loss": 0.4209,
+      "step": 824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811228857316056e-05,
+      "loss": 1.0654,
+      "step": 825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805385789485003e-05,
+      "loss": 1.1497,
+      "step": 826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779953589395947e-05,
+      "loss": 1.1315,
+      "step": 827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77936791758566e-05,
+      "loss": 1.1128,
+      "step": 828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787815640299514e-05,
+      "loss": 1.1004,
+      "step": 829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781945292417284e-05,
+      "loss": 1.0856,
+      "step": 830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777606813734495e-05,
+      "loss": 1.1036,
+      "step": 831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770184180223496e-05,
+      "loss": 1.1299,
+      "step": 832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764293426199873e-05,
+      "loss": 1.0871,
+      "step": 833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758395880426963e-05,
+      "loss": 1.1083,
+      "step": 834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752491548063594e-05,
+      "loss": 1.1526,
+      "step": 835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746580434274526e-05,
+      "loss": 1.0712,
+      "step": 836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740662544230463e-05,
+      "loss": 1.074,
+      "step": 837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734737883108026e-05,
+      "loss": 1.0564,
+      "step": 838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772880645608976e-05,
+      "loss": 1.0522,
+      "step": 839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772286826836413e-05,
+      "loss": 1.1151,
+      "step": 840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716923325125515e-05,
+      "loss": 1.0907,
+      "step": 841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7710971631574207e-05,
+      "loss": 1.095,
+      "step": 842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705013192916395e-05,
+      "loss": 1.056,
+      "step": 843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699048014364174e-05,
+      "loss": 1.1205,
+      "step": 844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769307610113553e-05,
+      "loss": 1.0923,
+      "step": 845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687097458454347e-05,
+      "loss": 1.0747,
+      "step": 846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681112091550393e-05,
+      "loss": 1.144,
+      "step": 847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767512000565931e-05,
+      "loss": 1.0651,
+      "step": 848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7669121206022625e-05,
+      "loss": 1.0979,
+      "step": 849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663115697887743e-05,
+      "loss": 1.1082,
+      "step": 850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657103486507924e-05,
+      "loss": 1.1294,
+      "step": 851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651084577142305e-05,
+      "loss": 1.0338,
+      "step": 852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645058975055872e-05,
+      "loss": 1.1006,
+      "step": 853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639026685519464e-05,
+      "loss": 1.0946,
+      "step": 854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632987713809782e-05,
+      "loss": 0.5315,
+      "step": 855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626942065209364e-05,
+      "loss": 1.085,
+      "step": 856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620889745006588e-05,
+      "loss": 1.0968,
+      "step": 857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614830758495668e-05,
+      "loss": 1.037,
+      "step": 858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760876511097665e-05,
+      "loss": 1.1184,
+      "step": 859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760269280775541e-05,
+      "loss": 1.0619,
+      "step": 860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7596613854143646e-05,
+      "loss": 1.0879,
+      "step": 861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590528255458862e-05,
+      "loss": 0.4699,
+      "step": 862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584436017024392e-05,
+      "loss": 1.1232,
+      "step": 863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7578337144169367e-05,
+      "loss": 1.1255,
+      "step": 864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757223164222872e-05,
+      "loss": 1.1352,
+      "step": 865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566119516543188e-05,
+      "loss": 1.1294,
+      "step": 866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560000772459307e-05,
+      "loss": 1.1178,
+      "step": 867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553875415329388e-05,
+      "loss": 1.1002,
+      "step": 868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547743450511538e-05,
+      "loss": 1.0915,
+      "step": 869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541604883369643e-05,
+      "loss": 1.0964,
+      "step": 870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753545971927336e-05,
+      "loss": 1.1086,
+      "step": 871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529307963598116e-05,
+      "loss": 0.4809,
+      "step": 872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7523149621725113e-05,
+      "loss": 1.1236,
+      "step": 873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75169846990413e-05,
+      "loss": 1.1215,
+      "step": 874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75108132009394e-05,
+      "loss": 1.0888,
+      "step": 875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504635132817867e-05,
+      "loss": 0.4013,
+      "step": 876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498450500080925e-05,
+      "loss": 1.1188,
+      "step": 877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749225930813852e-05,
+      "loss": 1.1222,
+      "step": 878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486061562406348e-05,
+      "loss": 1.0783,
+      "step": 879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479857268305832e-05,
+      "loss": 1.1708,
+      "step": 880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473646431264125e-05,
+      "loss": 1.0855,
+      "step": 881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467429056714108e-05,
+      "loss": 1.1331,
+      "step": 882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746120515009437e-05,
+      "loss": 1.0956,
+      "step": 883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454974716849225e-05,
+      "loss": 1.0696,
+      "step": 884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744873776242869e-05,
+      "loss": 1.1315,
+      "step": 885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442494292288486e-05,
+      "loss": 1.0769,
+      "step": 886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436244311890035e-05,
+      "loss": 1.1139,
+      "step": 887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429987826700455e-05,
+      "loss": 0.3825,
+      "step": 888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423724842192557e-05,
+      "loss": 1.1717,
+      "step": 889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417455363844823e-05,
+      "loss": 1.1044,
+      "step": 890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411179397141436e-05,
+      "loss": 1.0592,
+      "step": 891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404896947572237e-05,
+      "loss": 1.0943,
+      "step": 892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7398608020632752e-05,
+      "loss": 1.1043,
+      "step": 893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392312621824158e-05,
+      "loss": 1.0911,
+      "step": 894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386010756653305e-05,
+      "loss": 1.1245,
+      "step": 895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73797024306327e-05,
+      "loss": 1.1137,
+      "step": 896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737338764928049e-05,
+      "loss": 1.1052,
+      "step": 897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367066418120483e-05,
+      "loss": 1.1221,
+      "step": 898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360738742682115e-05,
+      "loss": 1.1004,
+      "step": 899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354404628500468e-05,
+      "loss": 1.1338,
+      "step": 900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348064081116258e-05,
+      "loss": 1.0659,
+      "step": 901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734171710607582e-05,
+      "loss": 1.0481,
+      "step": 902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733536370893112e-05,
+      "loss": 1.0758,
+      "step": 903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329003895239735e-05,
+      "loss": 1.1112,
+      "step": 904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322637670564858e-05,
+      "loss": 0.4059,
+      "step": 905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316265040475286e-05,
+      "loss": 1.0965,
+      "step": 906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730988601054543e-05,
+      "loss": 1.1362,
+      "step": 907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730350058635529e-05,
+      "loss": 1.1045,
+      "step": 908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297108773490463e-05,
+      "loss": 1.1003,
+      "step": 909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7290710577542125e-05,
+      "loss": 1.0977,
+      "step": 910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728430600410705e-05,
+      "loss": 1.1257,
+      "step": 911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727789505878758e-05,
+      "loss": 1.0706,
+      "step": 912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271477747191635e-05,
+      "loss": 1.1621,
+      "step": 913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7265054074932708e-05,
+      "loss": 1.0998,
+      "step": 914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258624047629845e-05,
+      "loss": 1.1,
+      "step": 915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725218767090766e-05,
+      "loss": 1.0944,
+      "step": 916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7245744950396324e-05,
+      "loss": 1.066,
+      "step": 917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7239295891731542e-05,
+      "loss": 1.0663,
+      "step": 918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723284050055458e-05,
+      "loss": 1.113,
+      "step": 919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226378782512233e-05,
+      "loss": 1.0945,
+      "step": 920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7219910743256834e-05,
+      "loss": 1.0375,
+      "step": 921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213436388446245e-05,
+      "loss": 1.1134,
+      "step": 922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206955723743856e-05,
+      "loss": 1.1268,
+      "step": 923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200468754818566e-05,
+      "loss": 1.0646,
+      "step": 924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71939754873448e-05,
+      "loss": 1.1257,
+      "step": 925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718747592700249e-05,
+      "loss": 1.0868,
+      "step": 926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180970079477066e-05,
+      "loss": 1.0653,
+      "step": 927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174457950459467e-05,
+      "loss": 1.122,
+      "step": 928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716793954564612e-05,
+      "loss": 1.0707,
+      "step": 929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161414870738946e-05,
+      "loss": 1.0819,
+      "step": 930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154883931445345e-05,
+      "loss": 0.4137,
+      "step": 931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71483467334782e-05,
+      "loss": 1.0676,
+      "step": 932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141803282555874e-05,
+      "loss": 1.0801,
+      "step": 933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135253584402193e-05,
+      "loss": 1.1084,
+      "step": 934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128697644746442e-05,
+      "loss": 1.1134,
+      "step": 935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122135469323384e-05,
+      "loss": 0.3793,
+      "step": 936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115567063873225e-05,
+      "loss": 1.0354,
+      "step": 937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710899243414161e-05,
+      "loss": 1.0796,
+      "step": 938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102411585879656e-05,
+      "loss": 1.0803,
+      "step": 939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709582452484389e-05,
+      "loss": 1.0776,
+      "step": 940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7089231256796294e-05,
+      "loss": 1.1172,
+      "step": 941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708263178750427e-05,
+      "loss": 1.1133,
+      "step": 942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076026122740653e-05,
+      "loss": 1.0806,
+      "step": 943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069414268283684e-05,
+      "loss": 1.0859,
+      "step": 944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7062796229917032e-05,
+      "loss": 1.0651,
+      "step": 945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705617201342977e-05,
+      "loss": 1.0904,
+      "step": 946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704954162461637e-05,
+      "loss": 1.0914,
+      "step": 947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704290506927671e-05,
+      "loss": 1.0537,
+      "step": 948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036262353216064e-05,
+      "loss": 1.1013,
+      "step": 949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029613482245082e-05,
+      "loss": 1.0896,
+      "step": 950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022958462179816e-05,
+      "loss": 1.1179,
+      "step": 951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701629729884168e-05,
+      "loss": 1.1506,
+      "step": 952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.700962999805748e-05,
+      "loss": 1.0844,
+      "step": 953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700295656565937e-05,
+      "loss": 1.1049,
+      "step": 954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996277007484884e-05,
+      "loss": 1.1161,
+      "step": 955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989591329376906e-05,
+      "loss": 0.3938,
+      "step": 956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982899537183674e-05,
+      "loss": 1.0469,
+      "step": 957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697620163675878e-05,
+      "loss": 1.0603,
+      "step": 958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6969497633961158e-05,
+      "loss": 0.4004,
+      "step": 959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696278753465507e-05,
+      "loss": 1.0911,
+      "step": 960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956071344710116e-05,
+      "loss": 1.1041,
+      "step": 961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949349070001237e-05,
+      "loss": 1.0952,
+      "step": 962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942620716408677e-05,
+      "loss": 1.0941,
+      "step": 963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935886289818005e-05,
+      "loss": 1.0805,
+      "step": 964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929145796120104e-05,
+      "loss": 1.0613,
+      "step": 965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922399241211168e-05,
+      "loss": 1.1223,
+      "step": 966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915646630992685e-05,
+      "loss": 1.1059,
+      "step": 967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908887971371446e-05,
+      "loss": 1.073,
+      "step": 968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902123268259524e-05,
+      "loss": 1.107,
+      "step": 969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895352527574288e-05,
+      "loss": 0.391,
+      "step": 970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688857575523839e-05,
+      "loss": 1.1301,
+      "step": 971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881792957179746e-05,
+      "loss": 1.0775,
+      "step": 972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6875004139331558e-05,
+      "loss": 1.1161,
+      "step": 973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686820930763228e-05,
+      "loss": 1.131,
+      "step": 974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861408468025642e-05,
+      "loss": 1.0492,
+      "step": 975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854601626460606e-05,
+      "loss": 1.0821,
+      "step": 976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847788788891412e-05,
+      "loss": 1.1264,
+      "step": 977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840969961277528e-05,
+      "loss": 1.0671,
+      "step": 978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683414514958366e-05,
+      "loss": 1.1315,
+      "step": 979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827314359779757e-05,
+      "loss": 1.0788,
+      "step": 980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820477597840997e-05,
+      "loss": 1.1013,
+      "step": 981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813634869747773e-05,
+      "loss": 1.0584,
+      "step": 982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806786181485704e-05,
+      "loss": 1.1201,
+      "step": 983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679993153904563e-05,
+      "loss": 1.1115,
+      "step": 984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793070948423582e-05,
+      "loss": 1.1154,
+      "step": 985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786204415620805e-05,
+      "loss": 0.3989,
+      "step": 986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.677933194664374e-05,
+      "loss": 1.122,
+      "step": 987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677245354750402e-05,
+      "loss": 1.1036,
+      "step": 988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765569224218464e-05,
+      "loss": 1.072,
+      "step": 989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675867898280908e-05,
+      "loss": 1.0285,
+      "step": 990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6751782829303045e-05,
+      "loss": 1.0826,
+      "step": 991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744880769732712e-05,
+      "loss": 1.1102,
+      "step": 992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673797281013559e-05,
+      "loss": 1.1134,
+      "step": 993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731058956554377e-05,
+      "loss": 1.0702,
+      "step": 994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6724139215036886e-05,
+      "loss": 1.0664,
+      "step": 995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6717213591636114e-05,
+      "loss": 1.0657,
+      "step": 996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710282092410182e-05,
+      "loss": 1.0726,
+      "step": 997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703344723422373e-05,
+      "loss": 1.0684,
+      "step": 998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696401490741082e-05,
+      "loss": 1.1374,
+      "step": 999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668945240043984e-05,
+      "loss": 1.1008,
+      "step": 1000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668249745859731e-05,
+      "loss": 1.0452,
+      "step": 1001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675536671297265e-05,
+      "loss": 1.1218,
+      "step": 1002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6668570044628597e-05,
+      "loss": 1.1262,
+      "step": 1003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661597584685303e-05,
+      "loss": 1.0368,
+      "step": 1004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665461929756648e-05,
+      "loss": 0.3936,
+      "step": 1005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664763518937632e-05,
+      "loss": 1.1745,
+      "step": 1006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640645266224127e-05,
+      "loss": 1.1113,
+      "step": 1007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633649534224266e-05,
+      "loss": 1.0814,
+      "step": 1008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662664799949619e-05,
+      "loss": 1.1107,
+      "step": 1009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661964066816444e-05,
+      "loss": 1.0989,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612627546358617e-05,
+      "loss": 0.3901,
+      "step": 1011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605608640213388e-05,
+      "loss": 1.0628,
+      "step": 1012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659858395586848e-05,
+      "loss": 1.0868,
+      "step": 1013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6591553499468678e-05,
+      "loss": 1.1061,
+      "step": 1014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584517277163814e-05,
+      "loss": 0.3975,
+      "step": 1015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657747529510876e-05,
+      "loss": 1.1186,
+      "step": 1016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6570427559463432e-05,
+      "loss": 1.086,
+      "step": 1017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6563374076392777e-05,
+      "loss": 1.099,
+      "step": 1018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6556314852066767e-05,
+      "loss": 1.0783,
+      "step": 1019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65492498926604e-05,
+      "loss": 1.0928,
+      "step": 1020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542179204353686e-05,
+      "loss": 1.0878,
+      "step": 1021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6535102793331654e-05,
+      "loss": 0.3594,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652802066578433e-05,
+      "loss": 1.1168,
+      "step": 1023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520932827906747e-05,
+      "loss": 1.0609,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651383928589893e-05,
+      "loss": 1.1305,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650674004596589e-05,
+      "loss": 1.1249,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499635114317633e-05,
+      "loss": 1.0839,
+      "step": 1027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492524497169133e-05,
+      "loss": 1.0259,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485408200740346e-05,
+      "loss": 1.0811,
+      "step": 1029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478286231256183e-05,
+      "loss": 1.029,
+      "step": 1030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647115859494653e-05,
+      "loss": 1.088,
+      "step": 1031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6464025298046234e-05,
+      "loss": 1.1112,
+      "step": 1032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456886346795075e-05,
+      "loss": 1.0718,
+      "step": 1033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449741747437795e-05,
+      "loss": 1.0932,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442591506224067e-05,
+      "loss": 1.0842,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435435629408503e-05,
+      "loss": 1.032,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428274123250655e-05,
+      "loss": 1.067,
+      "step": 1037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421106994014974e-05,
+      "loss": 1.0943,
+      "step": 1038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641393424797086e-05,
+      "loss": 1.1144,
+      "step": 1039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64067558913926e-05,
+      "loss": 1.0821,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399571930559407e-05,
+      "loss": 1.0497,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639238237175538e-05,
+      "loss": 1.1091,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385187221269537e-05,
+      "loss": 1.1125,
+      "step": 1043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637798648539577e-05,
+      "loss": 1.042,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370780170432852e-05,
+      "loss": 1.0477,
+      "step": 1045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636356828268446e-05,
+      "loss": 1.0314,
+      "step": 1046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635635082845912e-05,
+      "loss": 1.0723,
+      "step": 1047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634912781407024e-05,
+      "loss": 1.0792,
+      "step": 1048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341899245836094e-05,
+      "loss": 1.0881,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633466513007981e-05,
+      "loss": 1.0788,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632742547312937e-05,
+      "loss": 1.066,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320180281317594e-05,
+      "loss": 1.0943,
+      "step": 1052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631292956098216e-05,
+      "loss": 1.0869,
+      "step": 1053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305673318465565e-05,
+      "loss": 1.0628,
+      "step": 1054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298411560115156e-05,
+      "loss": 1.0837,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629114429228309e-05,
+      "loss": 1.1321,
+      "step": 1056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6283871521326342e-05,
+      "loss": 1.1072,
+      "step": 1057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276593253606713e-05,
+      "loss": 1.0399,
+      "step": 1058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269309495490803e-05,
+      "loss": 1.0888,
+      "step": 1059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262020253350013e-05,
+      "loss": 1.106,
+      "step": 1060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254725533560555e-05,
+      "loss": 1.0581,
+      "step": 1061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247425342503415e-05,
+      "loss": 1.1039,
+      "step": 1062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240119686564374e-05,
+      "loss": 1.1278,
+      "step": 1063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232808572133988e-05,
+      "loss": 1.0778,
+      "step": 1064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225492005607603e-05,
+      "loss": 1.0641,
+      "step": 1065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621816999338531e-05,
+      "loss": 1.0692,
+      "step": 1066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210842541871993e-05,
+      "loss": 1.0573,
+      "step": 1067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620350965747726e-05,
+      "loss": 1.0517,
+      "step": 1068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196171346615497e-05,
+      "loss": 1.0869,
+      "step": 1069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618882761570583e-05,
+      "loss": 1.0941,
+      "step": 1070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6181478471172123e-05,
+      "loss": 1.0602,
+      "step": 1071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617412391944298e-05,
+      "loss": 1.0906,
+      "step": 1072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166763966951723e-05,
+      "loss": 1.0665,
+      "step": 1073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615939862013642e-05,
+      "loss": 1.0859,
+      "step": 1074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152027885439838e-05,
+      "loss": 1.0776,
+      "step": 1075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144651769309466e-05,
+      "loss": 1.1023,
+      "step": 1076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137270278197496e-05,
+      "loss": 1.052,
+      "step": 1077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612988341856083e-05,
+      "loss": 1.0861,
+      "step": 1078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612249119686105e-05,
+      "loss": 1.1026,
+      "step": 1079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115093619564447e-05,
+      "loss": 1.0715,
+      "step": 1080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6107690693141993e-05,
+      "loss": 1.0275,
+      "step": 1081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610028242406932e-05,
+      "loss": 1.1205,
+      "step": 1082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092868818826762e-05,
+      "loss": 1.1079,
+      "step": 1083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60854498838993e-05,
+      "loss": 1.1036,
+      "step": 1084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078025625776587e-05,
+      "loss": 1.1022,
+      "step": 1085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070596050952922e-05,
+      "loss": 1.0113,
+      "step": 1086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063161165927272e-05,
+      "loss": 1.0984,
+      "step": 1087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055720977203237e-05,
+      "loss": 1.1027,
+      "step": 1088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604827549128905e-05,
+      "loss": 1.0808,
+      "step": 1089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040824714697595e-05,
+      "loss": 1.0636,
+      "step": 1090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603336865394637e-05,
+      "loss": 1.0576,
+      "step": 1091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025907315557504e-05,
+      "loss": 1.1003,
+      "step": 1092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601844070605774e-05,
+      "loss": 1.0969,
+      "step": 1093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6010968831978423e-05,
+      "loss": 1.0663,
+      "step": 1094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600349169985552e-05,
+      "loss": 1.0767,
+      "step": 1095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996009316229585e-05,
+      "loss": 1.0575,
+      "step": 1096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988521687645767e-05,
+      "loss": 1.091,
+      "step": 1097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981028820653812e-05,
+      "loss": 1.1132,
+      "step": 1098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597353072180803e-05,
+      "loss": 0.3818,
+      "step": 1099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596602739766733e-05,
+      "loss": 1.0899,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958518854795173e-05,
+      "loss": 1.0707,
+      "step": 1101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951005099759594e-05,
+      "loss": 1.0825,
+      "step": 1102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594348613913319e-05,
+      "loss": 1.0854,
+      "step": 1103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935961979493103e-05,
+      "loss": 1.0956,
+      "step": 1104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928432627421034e-05,
+      "loss": 1.0683,
+      "step": 1105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592089808950321e-05,
+      "loss": 1.1059,
+      "step": 1106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591335837233041e-05,
+      "loss": 1.0785,
+      "step": 1107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905813482497934e-05,
+      "loss": 1.0653,
+      "step": 1108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898263426605615e-05,
+      "loss": 1.0961,
+      "step": 1109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589070821125779e-05,
+      "loss": 1.0659,
+      "step": 1110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588314784306333e-05,
+      "loss": 1.1421,
+      "step": 1111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875582328635595e-05,
+      "loss": 1.0938,
+      "step": 1112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868011674592457e-05,
+      "loss": 1.0754,
+      "step": 1113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586043588755628e-05,
+      "loss": 0.402,
+      "step": 1114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852854974153923e-05,
+      "loss": 1.1013,
+      "step": 1115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584526894101672e-05,
+      "loss": 1.0912,
+      "step": 1116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837677794780487e-05,
+      "loss": 1.078,
+      "step": 1117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583008154208552e-05,
+      "loss": 1.0993,
+      "step": 1118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822480189576573e-05,
+      "loss": 1.1032,
+      "step": 1119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814873743902867e-05,
+      "loss": 1.0977,
+      "step": 1120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807262211718067e-05,
+      "loss": 1.0776,
+      "step": 1121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799645599680306e-05,
+      "loss": 1.0209,
+      "step": 1122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5792023914452145e-05,
+      "loss": 1.0514,
+      "step": 1123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578439716270059e-05,
+      "loss": 1.0814,
+      "step": 1124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776765351097076e-05,
+      "loss": 1.0716,
+      "step": 1125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5769128486317466e-05,
+      "loss": 1.0002,
+      "step": 1126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761486575042038e-05,
+      "loss": 1.1235,
+      "step": 1127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753839623955492e-05,
+      "loss": 1.0981,
+      "step": 1128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746187639746932e-05,
+      "loss": 1.0524,
+      "step": 1129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738530629109866e-05,
+      "loss": 1.0761,
+      "step": 1130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57308685987422e-05,
+      "loss": 1.0379,
+      "step": 1131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572320155534623e-05,
+      "loss": 1.1253,
+      "step": 1132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715529505628632e-05,
+      "loss": 1.0886,
+      "step": 1133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707852456300475e-05,
+      "loss": 1.0979,
+      "step": 1134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570017041407718e-05,
+      "loss": 1.1207,
+      "step": 1135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5692483385678556e-05,
+      "loss": 1.0971,
+      "step": 1136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684791377828765e-05,
+      "loss": 1.0402,
+      "step": 1137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677094397256317e-05,
+      "loss": 0.3952,
+      "step": 1138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5669392450694096e-05,
+      "loss": 1.1335,
+      "step": 1139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56616855448793e-05,
+      "loss": 1.0896,
+      "step": 1140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565397368655348e-05,
+      "loss": 1.0952,
+      "step": 1141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646256882462525e-05,
+      "loss": 1.0765,
+      "step": 1142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638535139356637e-05,
+      "loss": 1.0884,
+      "step": 1143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563080846399035e-05,
+      "loss": 1.0784,
+      "step": 1144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623076863122505e-05,
+      "loss": 1.0908,
+      "step": 1145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615340343516254e-05,
+      "loss": 1.0674,
+      "step": 1146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607598911939046e-05,
+      "loss": 1.1074,
+      "step": 1147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599852575162644e-05,
+      "loss": 1.0684,
+      "step": 1148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559210133996308e-05,
+      "loss": 0.3915,
+      "step": 1149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584345213120685e-05,
+      "loss": 1.114,
+      "step": 1150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576584201420056e-05,
+      "loss": 1.1023,
+      "step": 1151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568818311650088e-05,
+      "loss": 1.0798,
+      "step": 1152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561047550603912e-05,
+      "loss": 1.0474,
+      "step": 1153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553271925078942e-05,
+      "loss": 1.056,
+      "step": 1154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554549144187684e-05,
+      "loss": 1.099,
+      "step": 1155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553770610780351e-05,
+      "loss": 1.1312,
+      "step": 1156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552991592966912e-05,
+      "loss": 1.0729,
+      "step": 1157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5522120914288047e-05,
+      "loss": 1.0471,
+      "step": 1158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514321068478923e-05,
+      "loss": 1.0644,
+      "step": 1159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5506516399064593e-05,
+      "loss": 1.0624,
+      "step": 1160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5498706912872125e-05,
+      "loss": 1.0556,
+      "step": 1161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490892616732803e-05,
+      "loss": 1.0263,
+      "step": 1162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548307351748211e-05,
+      "loss": 1.0986,
+      "step": 1163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475249621959743e-05,
+      "loss": 1.0648,
+      "step": 1164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467420937009586e-05,
+      "loss": 1.0183,
+      "step": 1165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545958746947971e-05,
+      "loss": 1.0971,
+      "step": 1166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451749226222382e-05,
+      "loss": 1.0577,
+      "step": 1167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443906214094028e-05,
+      "loss": 1.0775,
+      "step": 1168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436058439955263e-05,
+      "loss": 1.0165,
+      "step": 1169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428205910670854e-05,
+      "loss": 1.0721,
+      "step": 1170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420348633109745e-05,
+      "loss": 1.0237,
+      "step": 1171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541248661414501e-05,
+      "loss": 1.0768,
+      "step": 1172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540461986065389e-05,
+      "loss": 1.1057,
+      "step": 1173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539674837951776e-05,
+      "loss": 1.0923,
+      "step": 1174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388872177622126e-05,
+      "loss": 1.0249,
+      "step": 1175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380991261856633e-05,
+      "loss": 1.0993,
+      "step": 1176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373105639115043e-05,
+      "loss": 1.0138,
+      "step": 1177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365215316295243e-05,
+      "loss": 1.0509,
+      "step": 1178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357320300299212e-05,
+      "loss": 1.115,
+      "step": 1179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349420598033054e-05,
+      "loss": 1.0578,
+      "step": 1180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341516216406976e-05,
+      "loss": 0.3944,
+      "step": 1181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333607162335254e-05,
+      "loss": 1.0914,
+      "step": 1182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325693442736273e-05,
+      "loss": 1.0641,
+      "step": 1183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531777506453249e-05,
+      "loss": 1.0331,
+      "step": 1184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530985203465044e-05,
+      "loss": 1.0907,
+      "step": 1185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530192436002072e-05,
+      "loss": 1.08,
+      "step": 1186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5293992047578e-05,
+      "loss": 1.0825,
+      "step": 1187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286055104261004e-05,
+      "loss": 1.0788,
+      "step": 1188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278113537012503e-05,
+      "loss": 1.0735,
+      "step": 1189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5270167352779317e-05,
+      "loss": 1.0497,
+      "step": 1190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52622165585123e-05,
+      "loss": 0.4086,
+      "step": 1191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254261161166342e-05,
+      "loss": 1.1267,
+      "step": 1192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246301167700361e-05,
+      "loss": 1.0806,
+      "step": 1193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238336585077286e-05,
+      "loss": 1.0607,
+      "step": 1194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5230367420264081e-05,
+      "loss": 1.0703,
+      "step": 1195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222393680231693e-05,
+      "loss": 1.0742,
+      "step": 1196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214415371955085e-05,
+      "loss": 1.1039,
+      "step": 1197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206432502413217e-05,
+      "loss": 1.0343,
+      "step": 1198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198445078589035e-05,
+      "loss": 1.1124,
+      "step": 1199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519045310746947e-05,
+      "loss": 1.0538,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518245659604543e-05,
+      "loss": 1.0705,
+      "step": 1201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174455551311792e-05,
+      "loss": 1.1009,
+      "step": 1202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166449980267403e-05,
+      "loss": 1.0695,
+      "step": 1203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158439889915075e-05,
+      "loss": 1.1628,
+      "step": 1204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150425287261556e-05,
+      "loss": 1.0693,
+      "step": 1205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142406179317555e-05,
+      "loss": 1.1009,
+      "step": 1206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134382573097721e-05,
+      "loss": 1.0873,
+      "step": 1207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126354475620629e-05,
+      "loss": 1.0912,
+      "step": 1208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118321893908792e-05,
+      "loss": 1.0748,
+      "step": 1209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110284834988641e-05,
+      "loss": 1.088,
+      "step": 1210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102243305890531e-05,
+      "loss": 1.0885,
+      "step": 1211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094197313648713e-05,
+      "loss": 1.0605,
+      "step": 1212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086146865301353e-05,
+      "loss": 1.1383,
+      "step": 1213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078091967890504e-05,
+      "loss": 1.0017,
+      "step": 1214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070032628462132e-05,
+      "loss": 0.4257,
+      "step": 1215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061968854066068e-05,
+      "loss": 1.05,
+      "step": 1216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505390065175603e-05,
+      "loss": 1.0717,
+      "step": 1217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045828028589612e-05,
+      "loss": 1.0506,
+      "step": 1218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503775099162827e-05,
+      "loss": 1.0608,
+      "step": 1219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029669547937322e-05,
+      "loss": 1.0415,
+      "step": 1220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021583704585943e-05,
+      "loss": 1.0303,
+      "step": 1221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013493468647153e-05,
+      "loss": 1.0725,
+      "step": 1222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005398847197819e-05,
+      "loss": 0.9935,
+      "step": 1223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997299847318643e-05,
+      "loss": 0.4188,
+      "step": 1224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989196476094153e-05,
+      "loss": 1.1212,
+      "step": 1225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981088740612705e-05,
+      "loss": 1.0504,
+      "step": 1226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972976647966472e-05,
+      "loss": 1.0547,
+      "step": 1227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964860205251438e-05,
+      "loss": 1.0061,
+      "step": 1228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495673941956739e-05,
+      "loss": 1.0874,
+      "step": 1229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4948614298017918e-05,
+      "loss": 1.1079,
+      "step": 1230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49404848477104e-05,
+      "loss": 1.0938,
+      "step": 1231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932351075756005e-05,
+      "loss": 1.1172,
+      "step": 1232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924212989269679e-05,
+      "loss": 0.3997,
+      "step": 1233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916070595370147e-05,
+      "loss": 1.0677,
+      "step": 1234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907923901179896e-05,
+      "loss": 1.0915,
+      "step": 1235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899772913825174e-05,
+      "loss": 1.0826,
+      "step": 1236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891617640435994e-05,
+      "loss": 1.1078,
+      "step": 1237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883458088146108e-05,
+      "loss": 1.0686,
+      "step": 1238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875294264093013e-05,
+      "loss": 1.0696,
+      "step": 1239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867126175417945e-05,
+      "loss": 0.3916,
+      "step": 1240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858953829265871e-05,
+      "loss": 1.0589,
+      "step": 1241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850777232785486e-05,
+      "loss": 1.094,
+      "step": 1242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484259639312919e-05,
+      "loss": 1.0685,
+      "step": 1243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834411317453103e-05,
+      "loss": 1.0815,
+      "step": 1244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826222012917055e-05,
+      "loss": 0.366,
+      "step": 1245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818028486684567e-05,
+      "loss": 1.0755,
+      "step": 1246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809830745922855e-05,
+      "loss": 1.0405,
+      "step": 1247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801628797802825e-05,
+      "loss": 1.087,
+      "step": 1248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793422649499057e-05,
+      "loss": 1.0541,
+      "step": 1249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785212308189816e-05,
+      "loss": 1.0624,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477699778105702e-05,
+      "loss": 1.1069,
+      "step": 1251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768779075286266e-05,
+      "loss": 1.0523,
+      "step": 1252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760556198066786e-05,
+      "loss": 1.0551,
+      "step": 1253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752329156591478e-05,
+      "loss": 0.3899,
+      "step": 1254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744097958056873e-05,
+      "loss": 1.0731,
+      "step": 1255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735862609663145e-05,
+      "loss": 1.0591,
+      "step": 1256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727623118614092e-05,
+      "loss": 1.0479,
+      "step": 1257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719379492117138e-05,
+      "loss": 1.0571,
+      "step": 1258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711131737383328e-05,
+      "loss": 1.1007,
+      "step": 1259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470287986162731e-05,
+      "loss": 1.0689,
+      "step": 1260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694623872067348e-05,
+      "loss": 1.088,
+      "step": 1261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686363775925295e-05,
+      "loss": 1.0673,
+      "step": 1262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678099580426596e-05,
+      "loss": 1.0389,
+      "step": 1263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4669831292800292e-05,
+      "loss": 0.3686,
+      "step": 1264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.466155892027899e-05,
+      "loss": 1.1016,
+      "step": 1265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653282470098881e-05,
+      "loss": 1.0546,
+      "step": 1266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464500194949972e-05,
+      "loss": 1.0522,
+      "step": 1267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463671736572482e-05,
+      "loss": 1.1104,
+      "step": 1268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628428726021052e-05,
+      "loss": 1.0928,
+      "step": 1269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620136037638828e-05,
+      "loss": 1.0645,
+      "step": 1270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611839307832112e-05,
+      "loss": 1.0869,
+      "step": 1271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460353854385839e-05,
+      "loss": 1.1099,
+      "step": 1272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595233752978694e-05,
+      "loss": 1.1326,
+      "step": 1273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586924942457561e-05,
+      "loss": 1.0699,
+      "step": 1274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578612119563053e-05,
+      "loss": 1.0968,
+      "step": 1275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4570295291566745e-05,
+      "loss": 1.1095,
+      "step": 1276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561974465743706e-05,
+      "loss": 1.0872,
+      "step": 1277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553649649372515e-05,
+      "loss": 1.093,
+      "step": 1278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545320849735222e-05,
+      "loss": 1.1328,
+      "step": 1279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536988074117386e-05,
+      "loss": 1.0693,
+      "step": 1280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452865132980802e-05,
+      "loss": 1.064,
+      "step": 1281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452031062409963e-05,
+      "loss": 1.0622,
+      "step": 1282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4511965964288169e-05,
+      "loss": 1.0572,
+      "step": 1283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503617357673058e-05,
+      "loss": 1.0855,
+      "step": 1284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495264811557174e-05,
+      "loss": 1.1164,
+      "step": 1285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486908333246831e-05,
+      "loss": 1.1156,
+      "step": 1286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478547930051789e-05,
+      "loss": 1.0755,
+      "step": 1287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447018360928524e-05,
+      "loss": 0.3712,
+      "step": 1288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44618153782638e-05,
+      "loss": 1.0816,
+      "step": 1289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453443244307508e-05,
+      "loss": 1.1183,
+      "step": 1290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445067214739816e-05,
+      "loss": 1.09,
+      "step": 1291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4436687296887584e-05,
+      "loss": 1.078,
+      "step": 1292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428303498081074e-05,
+      "loss": 1.1235,
+      "step": 1293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419915825653946e-05,
+      "loss": 1.046,
+      "step": 1294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411524286943241e-05,
+      "loss": 1.0784,
+      "step": 1295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403128889289384e-05,
+      "loss": 1.013,
+      "step": 1296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394729640036183e-05,
+      "loss": 1.0521,
+      "step": 1297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386326546530802e-05,
+      "loss": 1.0972,
+      "step": 1298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4377919616123784e-05,
+      "loss": 1.0801,
+      "step": 1299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436950885616901e-05,
+      "loss": 1.0708,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361094274023726e-05,
+      "loss": 1.0619,
+      "step": 1301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352675877048516e-05,
+      "loss": 1.035,
+      "step": 1302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43442536726073e-05,
+      "loss": 1.0809,
+      "step": 1303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335827668067328e-05,
+      "loss": 1.0408,
+      "step": 1304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327397870799175e-05,
+      "loss": 1.0548,
+      "step": 1305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431896428817674e-05,
+      "loss": 1.1112,
+      "step": 1306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431052692757722e-05,
+      "loss": 1.0924,
+      "step": 1307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302085796381126e-05,
+      "loss": 1.1086,
+      "step": 1308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293640901972267e-05,
+      "loss": 0.3962,
+      "step": 1309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285192251737741e-05,
+      "loss": 1.0421,
+      "step": 1310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276739853067935e-05,
+      "loss": 1.0988,
+      "step": 1311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268283713356508e-05,
+      "loss": 1.0847,
+      "step": 1312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42598238400004e-05,
+      "loss": 1.0639,
+      "step": 1313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425136024039981e-05,
+      "loss": 0.3676,
+      "step": 1314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42428929219582e-05,
+      "loss": 1.0336,
+      "step": 1315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234421892082289e-05,
+      "loss": 1.0635,
+      "step": 1316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225947158182028e-05,
+      "loss": 1.089,
+      "step": 1317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4217468727670624e-05,
+      "loss": 1.0336,
+      "step": 1318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208986607964514e-05,
+      "loss": 1.0378,
+      "step": 1319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420050080648336e-05,
+      "loss": 1.0862,
+      "step": 1320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192011330650037e-05,
+      "loss": 1.0496,
+      "step": 1321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183518187890647e-05,
+      "loss": 1.0503,
+      "step": 1322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175021385634495e-05,
+      "loss": 1.016,
+      "step": 1323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166520931314081e-05,
+      "loss": 1.0604,
+      "step": 1324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158016832365111e-05,
+      "loss": 1.0907,
+      "step": 1325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149509096226468e-05,
+      "loss": 0.3878,
+      "step": 1326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414099773034022e-05,
+      "loss": 1.0897,
+      "step": 1327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4132482742151618e-05,
+      "loss": 1.0817,
+      "step": 1328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123964139109072e-05,
+      "loss": 1.12,
+      "step": 1329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4115441928664156e-05,
+      "loss": 1.0517,
+      "step": 1330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106916118271601e-05,
+      "loss": 1.0484,
+      "step": 1331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4098386715389288e-05,
+      "loss": 1.0838,
+      "step": 1332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089853727478238e-05,
+      "loss": 0.3992,
+      "step": 1333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408131716200261e-05,
+      "loss": 1.0513,
+      "step": 1334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072777026429688e-05,
+      "loss": 1.104,
+      "step": 1335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406423332822989e-05,
+      "loss": 1.072,
+      "step": 1336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405568607487674e-05,
+      "loss": 1.0868,
+      "step": 1337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404713527384687e-05,
+      "loss": 1.0466,
+      "step": 1338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038580932620028e-05,
+      "loss": 1.0658,
+      "step": 1339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030023058679043e-05,
+      "loss": 1.0859,
+      "step": 1340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021461659509847e-05,
+      "loss": 1.0951,
+      "step": 1341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012896742601452e-05,
+      "loss": 1.1033,
+      "step": 1342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004328315445941e-05,
+      "loss": 1.0998,
+      "step": 1343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995756385538473e-05,
+      "loss": 1.1083,
+      "step": 1344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987180960377273e-05,
+      "loss": 1.0591,
+      "step": 1345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397860204746362e-05,
+      "loss": 1.1314,
+      "step": 1346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970019654301847e-05,
+      "loss": 0.3668,
+      "step": 1347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961433788399323e-05,
+      "loss": 1.0778,
+      "step": 1348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3952844457266468e-05,
+      "loss": 1.1229,
+      "step": 1349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394425166841672e-05,
+      "loss": 0.3733,
+      "step": 1350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393565542936655e-05,
+      "loss": 1.0574,
+      "step": 1351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927055747635445e-05,
+      "loss": 1.0942,
+      "step": 1352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3918452630745902e-05,
+      "loss": 1.0696,
+      "step": 1353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909846086223427e-05,
+      "loss": 1.0613,
+      "step": 1354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901236121596515e-05,
+      "loss": 1.0863,
+      "step": 1355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892622744396667e-05,
+      "loss": 1.078,
+      "step": 1356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884005962158353e-05,
+      "loss": 1.0913,
+      "step": 1357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875385782419038e-05,
+      "loss": 1.064,
+      "step": 1358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386676221271914e-05,
+      "loss": 1.0862,
+      "step": 1359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858135260602064e-05,
+      "loss": 1.0747,
+      "step": 1360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849504933614152e-05,
+      "loss": 1.0881,
+      "step": 1361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840871239304718e-05,
+      "loss": 1.0635,
+      "step": 1362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832234185226008e-05,
+      "loss": 1.0588,
+      "step": 1363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823593778933211e-05,
+      "loss": 1.0538,
+      "step": 1364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3814950027984451e-05,
+      "loss": 1.0457,
+      "step": 1365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380630293994077e-05,
+      "loss": 1.0711,
+      "step": 1366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379765252236614e-05,
+      "loss": 1.1189,
+      "step": 1367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3788998782827439e-05,
+      "loss": 1.0821,
+      "step": 1368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3780341728894446e-05,
+      "loss": 1.1018,
+      "step": 1369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377168136813985e-05,
+      "loss": 1.0757,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763017708139227e-05,
+      "loss": 1.1312,
+      "step": 1371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754350756471042e-05,
+      "loss": 1.0136,
+      "step": 1372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745680520716633e-05,
+      "loss": 1.0984,
+      "step": 1373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737007008460212e-05,
+      "loss": 1.1078,
+      "step": 1374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728330227288868e-05,
+      "loss": 1.1084,
+      "step": 1375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719650184792533e-05,
+      "loss": 1.1326,
+      "step": 1376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3710966888564006e-05,
+      "loss": 1.0845,
+      "step": 1377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370228034619892e-05,
+      "loss": 1.0688,
+      "step": 1378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693590565295759e-05,
+      "loss": 1.0973,
+      "step": 1379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684897553455834e-05,
+      "loss": 1.0201,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676201318283278e-05,
+      "loss": 1.097,
+      "step": 1381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667501867385054e-05,
+      "loss": 1.041,
+      "step": 1382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365879920837093e-05,
+      "loss": 1.0913,
+      "step": 1383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650093348853483e-05,
+      "loss": 1.0885,
+      "step": 1384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641384296448088e-05,
+      "loss": 1.0519,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632672058772915e-05,
+      "loss": 1.0606,
+      "step": 1386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623956643448919e-05,
+      "loss": 1.0259,
+      "step": 1387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3615238058099835e-05,
+      "loss": 1.0156,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606516310352174e-05,
+      "loss": 1.0183,
+      "step": 1389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597791407835206e-05,
+      "loss": 1.0657,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589063358180967e-05,
+      "loss": 1.0651,
+      "step": 1391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580332169024244e-05,
+      "loss": 1.0331,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571597848002569e-05,
+      "loss": 1.114,
+      "step": 1393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562860402756215e-05,
+      "loss": 1.1264,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554119840928186e-05,
+      "loss": 1.0669,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545376170164215e-05,
+      "loss": 1.0627,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536629398112756e-05,
+      "loss": 1.0648,
+      "step": 1397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352787953242497e-05,
+      "loss": 1.0731,
+      "step": 1398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519126580754729e-05,
+      "loss": 1.0697,
+      "step": 1399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510370550758598e-05,
+      "loss": 1.0935,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501611450095844e-05,
+      "loss": 1.0767,
+      "step": 1401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492849286428414e-05,
+      "loss": 1.0648,
+      "step": 1402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3484084067420939e-05,
+      "loss": 1.0281,
+      "step": 1403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475315800740712e-05,
+      "loss": 1.0814,
+      "step": 1404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3466544494057705e-05,
+      "loss": 1.0506,
+      "step": 1405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3457770155044542e-05,
+      "loss": 1.115,
+      "step": 1406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448992791376502e-05,
+      "loss": 1.107,
+      "step": 1407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344021241073151e-05,
+      "loss": 1.1116,
+      "step": 1408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431429020790123e-05,
+      "loss": 1.066,
+      "step": 1409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422642629235542e-05,
+      "loss": 1.0548,
+      "step": 1410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413853243753587e-05,
+      "loss": 1.0433,
+      "step": 1411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3405060872032696e-05,
+      "loss": 1.0799,
+      "step": 1412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339626552176392e-05,
+      "loss": 0.9917,
+      "step": 1413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338746720064092e-05,
+      "loss": 1.0803,
+      "step": 1414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378665916359952e-05,
+      "loss": 1.1166,
+      "step": 1415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336986167661986e-05,
+      "loss": 1.1275,
+      "step": 1416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361054489122081e-05,
+      "loss": 1.0043,
+      "step": 1417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352244361570628e-05,
+      "loss": 1.0642,
+      "step": 1418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343431301672078e-05,
+      "loss": 1.0602,
+      "step": 1419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3334615317135589e-05,
+      "loss": 1.0943,
+      "step": 1420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332579641567286e-05,
+      "loss": 1.0737,
+      "step": 1421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316974604998153e-05,
+      "loss": 0.3991,
+      "step": 1422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308149892828274e-05,
+      "loss": 1.1033,
+      "step": 1423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3299322286882563e-05,
+      "loss": 1.0564,
+      "step": 1424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290491794882894e-05,
+      "loss": 1.0725,
+      "step": 1425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281658424553659e-05,
+      "loss": 1.0715,
+      "step": 1426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3272822183621777e-05,
+      "loss": 1.0882,
+      "step": 1427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263983079816674e-05,
+      "loss": 1.1216,
+      "step": 1428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325514112087028e-05,
+      "loss": 1.0319,
+      "step": 1429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246296314517024e-05,
+      "loss": 1.0836,
+      "step": 1430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237448668493822e-05,
+      "loss": 1.0863,
+      "step": 1431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322859819054008e-05,
+      "loss": 1.0579,
+      "step": 1432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219744888397676e-05,
+      "loss": 0.369,
+      "step": 1433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210888769810964e-05,
+      "loss": 1.095,
+      "step": 1434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202029842526752e-05,
+      "loss": 1.0602,
+      "step": 1435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193168114294315e-05,
+      "loss": 1.0782,
+      "step": 1436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318430359286537e-05,
+      "loss": 1.1006,
+      "step": 1437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.317543628599409e-05,
+      "loss": 1.0646,
+      "step": 1438
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166566201437065e-05,
+      "loss": 1.0506,
+      "step": 1439
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315769334695333e-05,
+      "loss": 1.0663,
+      "step": 1440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148817730304345e-05,
+      "loss": 1.0853,
+      "step": 1441
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139939359253969e-05,
+      "loss": 1.0915,
+      "step": 1442
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131058241568488e-05,
+      "loss": 1.0105,
+      "step": 1443
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122174385016583e-05,
+      "loss": 1.0636,
+      "step": 1444
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311328779736933e-05,
+      "loss": 1.0554,
+      "step": 1445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104398486400197e-05,
+      "loss": 1.0867,
+      "step": 1446
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095506459885031e-05,
+      "loss": 1.0569,
+      "step": 1447
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086611725602054e-05,
+      "loss": 0.9981,
+      "step": 1448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077714291331863e-05,
+      "loss": 1.0546,
+      "step": 1449
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068814164857414e-05,
+      "loss": 1.0746,
+      "step": 1450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059911353964015e-05,
+      "loss": 1.0718,
+      "step": 1451
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3051005866439317e-05,
+      "loss": 1.0869,
+      "step": 1452
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042097710073329e-05,
+      "loss": 0.3985,
+      "step": 1453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3033186892658378e-05,
+      "loss": 1.0234,
+      "step": 1454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302427342198913e-05,
+      "loss": 1.0567,
+      "step": 1455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3015357305862562e-05,
+      "loss": 0.3815,
+      "step": 1456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006438552077974e-05,
+      "loss": 1.0597,
+      "step": 1457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299751716843697e-05,
+      "loss": 1.0876,
+      "step": 1458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988593162743453e-05,
+      "loss": 1.0009,
+      "step": 1459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297966654280362e-05,
+      "loss": 1.0852,
+      "step": 1460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970737316425955e-05,
+      "loss": 1.0643,
+      "step": 1461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961805491421225e-05,
+      "loss": 1.0394,
+      "step": 1462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952871075602464e-05,
+      "loss": 1.0634,
+      "step": 1463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943934076784981e-05,
+      "loss": 1.0857,
+      "step": 1464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934994502786332e-05,
+      "loss": 1.075,
+      "step": 1465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292605236142634e-05,
+      "loss": 1.0366,
+      "step": 1466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917107660527064e-05,
+      "loss": 1.0699,
+      "step": 1467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908160407912802e-05,
+      "loss": 1.0797,
+      "step": 1468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2899210611410092e-05,
+      "loss": 1.0891,
+      "step": 1469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890258278847689e-05,
+      "loss": 1.0874,
+      "step": 1470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881303418056565e-05,
+      "loss": 1.1022,
+      "step": 1471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2872346036869919e-05,
+      "loss": 1.0482,
+      "step": 1472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2863386143123131e-05,
+      "loss": 1.105,
+      "step": 1473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28544237446538e-05,
+      "loss": 1.0264,
+      "step": 1474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2845458849301702e-05,
+      "loss": 1.0461,
+      "step": 1475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836491464908809e-05,
+      "loss": 1.0158,
+      "step": 1476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827521599319259e-05,
+      "loss": 1.076,
+      "step": 1477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818549260379366e-05,
+      "loss": 1.0315,
+      "step": 1478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280957445593761e-05,
+      "loss": 1.1121,
+      "step": 1479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2800597193844619e-05,
+      "loss": 1.1,
+      "step": 1480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279161748195318e-05,
+      "loss": 1.0748,
+      "step": 1481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782635328118217e-05,
+      "loss": 1.081,
+      "step": 1482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2773650740196793e-05,
+      "loss": 1.1124,
+      "step": 1483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764663726048101e-05,
+      "loss": 1.063,
+      "step": 1484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755674293533452e-05,
+      "loss": 1.0936,
+      "step": 1485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274668245051628e-05,
+      "loss": 1.0526,
+      "step": 1486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2737688204862115e-05,
+      "loss": 1.0462,
+      "step": 1487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2728691564438606e-05,
+      "loss": 1.0682,
+      "step": 1488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719692537115478e-05,
+      "loss": 1.0158,
+      "step": 1489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271069113076456e-05,
+      "loss": 1.0731,
+      "step": 1490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701687353259753e-05,
+      "loss": 0.3802,
+      "step": 1491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692681212477033e-05,
+      "loss": 1.0763,
+      "step": 1492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683672716294449e-05,
+      "loss": 0.9974,
+      "step": 1493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2674661872592098e-05,
+      "loss": 1.0925,
+      "step": 1494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665648689252148e-05,
+      "loss": 1.0764,
+      "step": 1495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656633174158799e-05,
+      "loss": 0.3452,
+      "step": 1496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2647615335198293e-05,
+      "loss": 1.0532,
+      "step": 1497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2638595180258912e-05,
+      "loss": 1.0819,
+      "step": 1498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2629572717230957e-05,
+      "loss": 1.0553,
+      "step": 1499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620547954006751e-05,
+      "loss": 1.0743,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2611520898480633e-05,
+      "loss": 1.062,
+      "step": 1501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602491558548937e-05,
+      "loss": 1.0736,
+      "step": 1502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2593459942110002e-05,
+      "loss": 1.0702,
+      "step": 1503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584426057064158e-05,
+      "loss": 1.0525,
+      "step": 1504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257538991131372e-05,
+      "loss": 1.064,
+      "step": 1505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566351512762973e-05,
+      "loss": 1.0983,
+      "step": 1506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2557310869318187e-05,
+      "loss": 1.0531,
+      "step": 1507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2548267988887584e-05,
+      "loss": 1.0697,
+      "step": 1508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539222879381346e-05,
+      "loss": 1.0709,
+      "step": 1509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2530175548711605e-05,
+      "loss": 1.0359,
+      "step": 1510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521126004792437e-05,
+      "loss": 1.0508,
+      "step": 1511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512074255539854e-05,
+      "loss": 1.0383,
+      "step": 1512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2503020308871794e-05,
+      "loss": 1.0159,
+      "step": 1513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493964172708121e-05,
+      "loss": 1.0507,
+      "step": 1514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248490585497061e-05,
+      "loss": 1.0529,
+      "step": 1515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2475845363582953e-05,
+      "loss": 1.0489,
+      "step": 1516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466782706470733e-05,
+      "loss": 1.0795,
+      "step": 1517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457717891561431e-05,
+      "loss": 0.3767,
+      "step": 1518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448650926784418e-05,
+      "loss": 1.049,
+      "step": 1519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2439581820070945e-05,
+      "loss": 1.023,
+      "step": 1520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430510579354134e-05,
+      "loss": 1.0844,
+      "step": 1521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421437212568977e-05,
+      "loss": 1.0407,
+      "step": 1522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241236172765232e-05,
+      "loss": 1.0462,
+      "step": 1523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2403284132542868e-05,
+      "loss": 1.0589,
+      "step": 1524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239420443518117e-05,
+      "loss": 1.0717,
+      "step": 1525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2385122643509615e-05,
+      "loss": 1.0997,
+      "step": 1526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237603876547242e-05,
+      "loss": 1.0404,
+      "step": 1527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366952809015633e-05,
+      "loss": 1.0831,
+      "step": 1528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357864782087117e-05,
+      "loss": 1.0598,
+      "step": 1529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348774692636539e-05,
+      "loss": 0.3905,
+      "step": 1530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339682548615383e-05,
+      "loss": 1.0982,
+      "step": 1531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2330588357976922e-05,
+      "loss": 1.0612,
+      "step": 1532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321492128676223e-05,
+      "loss": 1.0687,
+      "step": 1533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312393868670135e-05,
+      "loss": 1.0805,
+      "step": 1534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230329358591728e-05,
+      "loss": 1.065,
+      "step": 1535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229419128837806e-05,
+      "loss": 1.0104,
+      "step": 1536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228508698401462e-05,
+      "loss": 1.0014,
+      "step": 1537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275980680790884e-05,
+      "loss": 1.0458,
+      "step": 1538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266872386672507e-05,
+      "loss": 1.0451,
+      "step": 1539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225776210962689e-05,
+      "loss": 1.0841,
+      "step": 1540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2248649857623174e-05,
+      "loss": 1.0548,
+      "step": 1541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2239535638632221e-05,
+      "loss": 1.0897,
+      "step": 1542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223041946062662e-05,
+      "loss": 1.057,
+      "step": 1543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221301331580661e-05,
+      "loss": 1.045,
+      "step": 1544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212181259470362e-05,
+      "loss": 1.0466,
+      "step": 1545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203059252273416e-05,
+      "loss": 1.0342,
+      "step": 1546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219393531796923e-05,
+      "loss": 1.1463,
+      "step": 1547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184809464538883e-05,
+      "loss": 1.0046,
+      "step": 1548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2175681699965136e-05,
+      "loss": 1.0335,
+      "step": 1549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166552032232424e-05,
+      "loss": 1.0191,
+      "step": 1550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157420469326847e-05,
+      "loss": 1.0853,
+      "step": 1551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2148287019236165e-05,
+      "loss": 1.07,
+      "step": 1552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139151689949777e-05,
+      "loss": 1.0649,
+      "step": 1553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130014489458743e-05,
+      "loss": 1.0479,
+      "step": 1554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120875425755747e-05,
+      "loss": 1.0508,
+      "step": 1555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2111734506835103e-05,
+      "loss": 0.9894,
+      "step": 1556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102591740692757e-05,
+      "loss": 1.0047,
+      "step": 1557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093447135326263e-05,
+      "loss": 1.0341,
+      "step": 1558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084300698734787e-05,
+      "loss": 1.0729,
+      "step": 1559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075152438919095e-05,
+      "loss": 1.036,
+      "step": 1560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066002363881554e-05,
+      "loss": 1.0435,
+      "step": 1561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056850481626111e-05,
+      "loss": 1.0709,
+      "step": 1562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047696800158294e-05,
+      "loss": 1.024,
+      "step": 1563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038541327485212e-05,
+      "loss": 1.0354,
+      "step": 1564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029384071615528e-05,
+      "loss": 1.0567,
+      "step": 1565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020225040559485e-05,
+      "loss": 1.0941,
+      "step": 1566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2011064242328861e-05,
+      "loss": 1.0863,
+      "step": 1567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001901684936983e-05,
+      "loss": 1.0282,
+      "step": 1568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992737376398727e-05,
+      "loss": 1.0654,
+      "step": 1569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1983571324730488e-05,
+      "loss": 0.387,
+      "step": 1570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974403537950193e-05,
+      "loss": 1.0414,
+      "step": 1571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965234024077284e-05,
+      "loss": 1.0726,
+      "step": 1572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956062791132718e-05,
+      "loss": 0.9936,
+      "step": 1573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946889847138952e-05,
+      "loss": 1.0649,
+      "step": 1574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1937715200119935e-05,
+      "loss": 1.0371,
+      "step": 1575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1928538858101115e-05,
+      "loss": 0.9959,
+      "step": 1576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1919360829109419e-05,
+      "loss": 1.047,
+      "step": 1577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1910181121173252e-05,
+      "loss": 1.0436,
+      "step": 1578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190099974232248e-05,
+      "loss": 1.0249,
+      "step": 1579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891816700588439e-05,
+      "loss": 1.0795,
+      "step": 1580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1882632004003914e-05,
+      "loss": 1.0552,
+      "step": 1581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187344566060314e-05,
+      "loss": 1.0894,
+      "step": 1582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864257678421796e-05,
+      "loss": 1.0325,
+      "step": 1583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1855068065496984e-05,
+      "loss": 1.0746,
+      "step": 1584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845876829867242e-05,
+      "loss": 1.065,
+      "step": 1585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183668397957253e-05,
+      "loss": 1.0292,
+      "step": 1586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827489522654209e-05,
+      "loss": 0.9804,
+      "step": 1587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818293467155058e-05,
+      "loss": 1.0389,
+      "step": 1588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809095821119238e-05,
+      "loss": 1.0612,
+      "step": 1589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799896592592323e-05,
+      "loss": 1.062,
+      "step": 1590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790695789621249e-05,
+      "loss": 1.0535,
+      "step": 1591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781493420254346e-05,
+      "loss": 1.0373,
+      "step": 1592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772289492541304e-05,
+      "loss": 1.061,
+      "step": 1593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763084014533183e-05,
+      "loss": 0.3789,
+      "step": 1594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753876994282399e-05,
+      "loss": 1.0961,
+      "step": 1595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1744668439842705e-05,
+      "loss": 1.0508,
+      "step": 1596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735458359269216e-05,
+      "loss": 1.0496,
+      "step": 1597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726246760618364e-05,
+      "loss": 1.1056,
+      "step": 1598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717033651947917e-05,
+      "loss": 1.039,
+      "step": 1599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707819041316966e-05,
+      "loss": 0.9952,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169860293678591e-05,
+      "loss": 1.0772,
+      "step": 1601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689385346416458e-05,
+      "loss": 0.408,
+      "step": 1602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680166278271618e-05,
+      "loss": 1.0238,
+      "step": 1603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670945740415696e-05,
+      "loss": 0.3708,
+      "step": 1604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661723740914273e-05,
+      "loss": 1.0093,
+      "step": 1605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652500287834217e-05,
+      "loss": 1.0389,
+      "step": 1606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643275389243665e-05,
+      "loss": 1.0705,
+      "step": 1607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634049053212014e-05,
+      "loss": 1.0135,
+      "step": 1608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624821287809927e-05,
+      "loss": 1.0534,
+      "step": 1609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1615592101109309e-05,
+      "loss": 1.0304,
+      "step": 1610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1606361501183314e-05,
+      "loss": 1.0742,
+      "step": 1611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597129496106337e-05,
+      "loss": 1.0567,
+      "step": 1612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1587896093953982e-05,
+      "loss": 1.0343,
+      "step": 1613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578661302803101e-05,
+      "loss": 1.0411,
+      "step": 1614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569425130731742e-05,
+      "loss": 1.039,
+      "step": 1615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156018758581917e-05,
+      "loss": 1.1021,
+      "step": 1616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550948676145845e-05,
+      "loss": 1.0292,
+      "step": 1617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154170840979343e-05,
+      "loss": 1.0427,
+      "step": 1618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532466794844763e-05,
+      "loss": 1.0828,
+      "step": 1619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152322383938387e-05,
+      "loss": 1.0804,
+      "step": 1620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151397955149595e-05,
+      "loss": 0.3922,
+      "step": 1621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504733939267361e-05,
+      "loss": 0.3692,
+      "step": 1622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1495487010785626e-05,
+      "loss": 1.0343,
+      "step": 1623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486238774139414e-05,
+      "loss": 1.0874,
+      "step": 1624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476989237418543e-05,
+      "loss": 0.379,
+      "step": 1625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467738408713969e-05,
+      "loss": 1.0622,
+      "step": 1626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458486296117765e-05,
+      "loss": 0.3855,
+      "step": 1627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449232907723147e-05,
+      "loss": 1.0367,
+      "step": 1628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439978251624432e-05,
+      "loss": 1.0713,
+      "step": 1629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430722335917059e-05,
+      "loss": 1.0499,
+      "step": 1630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421465168697552e-05,
+      "loss": 1.077,
+      "step": 1631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412206758063546e-05,
+      "loss": 1.0373,
+      "step": 1632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402947112113751e-05,
+      "loss": 1.0717,
+      "step": 1633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393686238947965e-05,
+      "loss": 1.028,
+      "step": 1634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384424146667061e-05,
+      "loss": 1.0247,
+      "step": 1635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137516084337297e-05,
+      "loss": 1.0458,
+      "step": 1636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365896337168686e-05,
+      "loss": 1.0855,
+      "step": 1637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356630636158265e-05,
+      "loss": 1.0814,
+      "step": 1638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347363748446794e-05,
+      "loss": 1.1054,
+      "step": 1639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1338095682140403e-05,
+      "loss": 1.0351,
+      "step": 1640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328826445346258e-05,
+      "loss": 1.0037,
+      "step": 1641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319556046172539e-05,
+      "loss": 1.0614,
+      "step": 1642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310284492728451e-05,
+      "loss": 0.396,
+      "step": 1643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301011793124209e-05,
+      "loss": 1.0979,
+      "step": 1644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1291737955471022e-05,
+      "loss": 0.3591,
+      "step": 1645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1282462987881104e-05,
+      "loss": 1.0645,
+      "step": 1646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1273186898467657e-05,
+      "loss": 1.0387,
+      "step": 1647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263909695344852e-05,
+      "loss": 1.0695,
+      "step": 1648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254631386627852e-05,
+      "loss": 1.0318,
+      "step": 1649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245351980432771e-05,
+      "loss": 1.0615,
+      "step": 1650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236071484876694e-05,
+      "loss": 1.0265,
+      "step": 1651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226789908077657e-05,
+      "loss": 1.0587,
+      "step": 1652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217507258154634e-05,
+      "loss": 1.0381,
+      "step": 1653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208223543227547e-05,
+      "loss": 1.0835,
+      "step": 1654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198938771417246e-05,
+      "loss": 1.0918,
+      "step": 1655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118965295084551e-05,
+      "loss": 1.022,
+      "step": 1656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180366089635025e-05,
+      "loss": 1.0601,
+      "step": 1657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171078195909395e-05,
+      "loss": 1.093,
+      "step": 1658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161789277793129e-05,
+      "loss": 1.045,
+      "step": 1659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152499343411622e-05,
+      "loss": 1.0266,
+      "step": 1660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143208400891172e-05,
+      "loss": 0.4024,
+      "step": 1661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133916458358946e-05,
+      "loss": 1.0704,
+      "step": 1662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124623523942995e-05,
+      "loss": 1.0522,
+      "step": 1663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1115329605772232e-05,
+      "loss": 1.0328,
+      "step": 1664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1106034711976432e-05,
+      "loss": 1.0259,
+      "step": 1665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096738850686226e-05,
+      "loss": 1.0442,
+      "step": 1666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1087442030033086e-05,
+      "loss": 1.0494,
+      "step": 1667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078144258149332e-05,
+      "loss": 1.0235,
+      "step": 1668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068845543168103e-05,
+      "loss": 1.0566,
+      "step": 1669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059545893223378e-05,
+      "loss": 1.07,
+      "step": 1670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105024531644994e-05,
+      "loss": 1.0761,
+      "step": 1671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104094382098339e-05,
+      "loss": 1.0376,
+      "step": 1672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031641414960135e-05,
+      "loss": 1.0326,
+      "step": 1673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022338106517371e-05,
+      "loss": 1.0641,
+      "step": 1674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013033903793094e-05,
+      "loss": 1.0656,
+      "step": 1675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003728814926066e-05,
+      "loss": 1.0213,
+      "step": 1676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994422848055844e-05,
+      "loss": 1.0509,
+      "step": 1677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0985116011322738e-05,
+      "loss": 1.0623,
+      "step": 1678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975808312867822e-05,
+      "loss": 1.0731,
+      "step": 1679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0966499760832926e-05,
+      "loss": 1.0581,
+      "step": 1680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0957190363360633e-05,
+      "loss": 0.3851,
+      "step": 1681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947880128594253e-05,
+      "loss": 1.0573,
+      "step": 1682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938569064677832e-05,
+      "loss": 1.0717,
+      "step": 1683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929257179756153e-05,
+      "loss": 1.0473,
+      "step": 1684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0919944481974695e-05,
+      "loss": 1.0256,
+      "step": 1685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091063097947967e-05,
+      "loss": 1.0308,
+      "step": 1686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901316680417984e-05,
+      "loss": 1.0249,
+      "step": 1687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892001592937235e-05,
+      "loss": 1.0454,
+      "step": 1688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088268572518572e-05,
+      "loss": 1.0594,
+      "step": 1689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873369085312415e-05,
+      "loss": 1.0488,
+      "step": 1690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0864051681466969e-05,
+      "loss": 1.0221,
+      "step": 1691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854733521799705e-05,
+      "loss": 1.0614,
+      "step": 1692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08454146144616e-05,
+      "loss": 1.0603,
+      "step": 1693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836094967604292e-05,
+      "loss": 0.3576,
+      "step": 1694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826774589380056e-05,
+      "loss": 1.0877,
+      "step": 1695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081745348794182e-05,
+      "loss": 1.0601,
+      "step": 1696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0808131671443131e-05,
+      "loss": 1.0397,
+      "step": 1697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798809148038176e-05,
+      "loss": 1.0331,
+      "step": 1698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789485925881748e-05,
+      "loss": 1.0461,
+      "step": 1699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780162013129256e-05,
+      "loss": 1.0055,
+      "step": 1700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770837417936717e-05,
+      "loss": 1.0678,
+      "step": 1701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761512148460733e-05,
+      "loss": 1.0486,
+      "step": 1702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752186212858512e-05,
+      "loss": 1.0743,
+      "step": 1703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742859619287834e-05,
+      "loss": 0.9907,
+      "step": 1704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073353237590705e-05,
+      "loss": 1.0663,
+      "step": 1705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724204490875094e-05,
+      "loss": 1.0718,
+      "step": 1706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714875972351453e-05,
+      "loss": 1.0389,
+      "step": 1707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070554682849617e-05,
+      "loss": 0.9964,
+      "step": 1708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696217067469828e-05,
+      "loss": 1.0578,
+      "step": 1709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068688669743356e-05,
+      "loss": 1.0442,
+      "step": 1710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677555726549029e-05,
+      "loss": 1.0512,
+      "step": 1711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0668224162978415e-05,
+      "loss": 0.9982,
+      "step": 1712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658892014884433e-05,
+      "loss": 1.0489,
+      "step": 1713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649559290430291e-05,
+      "loss": 1.0354,
+      "step": 1714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0640225997779715e-05,
+      "loss": 1.0401,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0630892145096923e-05,
+      "loss": 1.0543,
+      "step": 1716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621557740546617e-05,
+      "loss": 1.0229,
+      "step": 1717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0612222792293999e-05,
+      "loss": 1.015,
+      "step": 1718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602887308504721e-05,
+      "loss": 1.0769,
+      "step": 1719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059355129734493e-05,
+      "loss": 1.0471,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0584214766981213e-05,
+      "loss": 1.028,
+      "step": 1721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574877725580628e-05,
+      "loss": 1.0833,
+      "step": 1722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565540181310664e-05,
+      "loss": 1.0395,
+      "step": 1723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0556202142339263e-05,
+      "loss": 1.0486,
+      "step": 1724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0546863616834794e-05,
+      "loss": 1.0463,
+      "step": 1725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537524612966052e-05,
+      "loss": 1.0506,
+      "step": 1726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0528185138902254e-05,
+      "loss": 1.0274,
+      "step": 1727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518845202813018e-05,
+      "loss": 1.0453,
+      "step": 1728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509504812868379e-05,
+      "loss": 1.0835,
+      "step": 1729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500163977238762e-05,
+      "loss": 0.4081,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049082270409498e-05,
+      "loss": 1.0426,
+      "step": 1731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481481001608235e-05,
+      "loss": 1.0631,
+      "step": 1732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04721388779501e-05,
+      "loss": 1.0209,
+      "step": 1733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046279634129252e-05,
+      "loss": 1.0215,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04534533998078e-05,
+      "loss": 1.0648,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044411006166859e-05,
+      "loss": 1.0402,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0434766335047908e-05,
+      "loss": 1.0546,
+      "step": 1737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0425422228119087e-05,
+      "loss": 1.0327,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041607774905581e-05,
+      "loss": 1.0225,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040673290603208e-05,
+      "loss": 1.0457,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397387707222215e-05,
+      "loss": 1.0299,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388042160800858e-05,
+      "loss": 1.0498,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378696274942932e-05,
+      "loss": 0.9573,
+      "step": 1743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369350057823681e-05,
+      "loss": 1.043,
+      "step": 1744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360003517618624e-05,
+      "loss": 1.0713,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350656662503569e-05,
+      "loss": 1.0535,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341309500654595e-05,
+      "loss": 1.0299,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331962040248056e-05,
+      "loss": 0.9881,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032261428946056e-05,
+      "loss": 1.0487,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.031326625646897e-05,
+      "loss": 1.0595,
+      "step": 1750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303917949450404e-05,
+      "loss": 1.0136,
+      "step": 1751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294569376582208e-05,
+      "loss": 1.047,
+      "step": 1752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285220546041972e-05,
+      "loss": 1.0189,
+      "step": 1753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275871466007499e-05,
+      "loss": 1.0433,
+      "step": 1754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266522144656823e-05,
+      "loss": 1.0629,
+      "step": 1755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257172590168179e-05,
+      "loss": 1.007,
+      "step": 1756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247822810720007e-05,
+      "loss": 1.0491,
+      "step": 1757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238472814490951e-05,
+      "loss": 1.0308,
+      "step": 1758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229122609659841e-05,
+      "loss": 1.0725,
+      "step": 1759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021977220440569e-05,
+      "loss": 0.994,
+      "step": 1760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021042160690768e-05,
+      "loss": 0.9981,
+      "step": 1761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0201070825345168e-05,
+      "loss": 1.0557,
+      "step": 1762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019171986789767e-05,
+      "loss": 1.0429,
+      "step": 1763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018236874274486e-05,
+      "loss": 1.0543,
+      "step": 1764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017301745806655e-05,
+      "loss": 1.0469,
+      "step": 1765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163666022042695e-05,
+      "loss": 1.0064,
+      "step": 1766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015431444285339e-05,
+      "loss": 1.0582,
+      "step": 1767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144962728678844e-05,
+      "loss": 0.9874,
+      "step": 1768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135610887699393e-05,
+      "loss": 1.0196,
+      "step": 1769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126258928095476e-05,
+      "loss": 1.0321,
+      "step": 1770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011690685804764e-05,
+      "loss": 1.083,
+      "step": 1771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107554685736531e-05,
+      "loss": 1.0516,
+      "step": 1772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009820241934288e-05,
+      "loss": 1.0183,
+      "step": 1773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0088850067047502e-05,
+      "loss": 1.0266,
+      "step": 1774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079497637031288e-05,
+      "loss": 0.9985,
+      "step": 1775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070145137475194e-05,
+      "loss": 1.0576,
+      "step": 1776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060792576560243e-05,
+      "loss": 1.0335,
+      "step": 1777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051439962467504e-05,
+      "loss": 1.0444,
+      "step": 1778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042087303378099e-05,
+      "loss": 1.0095,
+      "step": 1779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032734607473184e-05,
+      "loss": 1.0139,
+      "step": 1780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023381882933954e-05,
+      "loss": 1.0676,
+      "step": 1781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0014029137941618e-05,
+      "loss": 1.0455,
+      "step": 1782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0004676380677417e-05,
+      "loss": 1.0293,
+      "step": 1783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.995323619322588e-06,
+      "loss": 0.9761,
+      "step": 1784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985970862058385e-06,
+      "loss": 1.0636,
+      "step": 1785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976618117066048e-06,
+      "loss": 1.0086,
+      "step": 1786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967265392526817e-06,
+      "loss": 0.99,
+      "step": 1787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957912696621903e-06,
+      "loss": 1.0317,
+      "step": 1788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948560037532498e-06,
+      "loss": 1.0409,
+      "step": 1789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.939207423439762e-06,
+      "loss": 0.9668,
+      "step": 1790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.929854862524808e-06,
+      "loss": 1.0696,
+      "step": 1791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920502362968715e-06,
+      "loss": 0.3869,
+      "step": 1792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911149932952501e-06,
+      "loss": 1.0283,
+      "step": 1793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901797580657124e-06,
+      "loss": 1.0785,
+      "step": 1794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892445314263474e-06,
+      "loss": 1.0508,
+      "step": 1795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88309314195236e-06,
+      "loss": 1.0582,
+      "step": 1796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873741071904527e-06,
+      "loss": 0.9947,
+      "step": 1797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864389112300612e-06,
+      "loss": 1.0408,
+      "step": 1798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85503727132116e-06,
+      "loss": 1.0541,
+      "step": 1799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845685557146614e-06,
+      "loss": 1.0036,
+      "step": 1800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836333977957307e-06,
+      "loss": 1.0468,
+      "step": 1801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.826982541933453e-06,
+      "loss": 1.0645,
+      "step": 1802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817631257255143e-06,
+      "loss": 1.0386,
+      "step": 1803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808280132102333e-06,
+      "loss": 1.0546,
+      "step": 1804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798929174654832e-06,
+      "loss": 0.3938,
+      "step": 1805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789578393092322e-06,
+      "loss": 1.0249,
+      "step": 1806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780227795594314e-06,
+      "loss": 1.0114,
+      "step": 1807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770877390340162e-06,
+      "loss": 1.0375,
+      "step": 1808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76152718550905e-06,
+      "loss": 1.0175,
+      "step": 1809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752177189279995e-06,
+      "loss": 1.0883,
+      "step": 1810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742827409831826e-06,
+      "loss": 1.0283,
+      "step": 1811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73347785534318e-06,
+      "loss": 1.021,
+      "step": 1812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724128533992505e-06,
+      "loss": 1.0531,
+      "step": 1813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714779453958033e-06,
+      "loss": 1.0239,
+      "step": 1814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705430623417791e-06,
+      "loss": 1.0613,
+      "step": 1815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.696082050549597e-06,
+      "loss": 1.0859,
+      "step": 1816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686733743531032e-06,
+      "loss": 1.0533,
+      "step": 1817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677385710539443e-06,
+      "loss": 1.0158,
+      "step": 1818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.668037959751946e-06,
+      "loss": 1.0577,
+      "step": 1819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658690499345405e-06,
+      "loss": 1.0271,
+      "step": 1820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649343337496433e-06,
+      "loss": 1.0438,
+      "step": 1821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.639996482381379e-06,
+      "loss": 1.0611,
+      "step": 1822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630649942176324e-06,
+      "loss": 1.056,
+      "step": 1823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62130372505707e-06,
+      "loss": 1.0509,
+      "step": 1824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611957839199146e-06,
+      "loss": 1.0758,
+      "step": 1825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602612292777787e-06,
+      "loss": 1.0846,
+      "step": 1826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.593267093967924e-06,
+      "loss": 1.0611,
+      "step": 1827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583922250944195e-06,
+      "loss": 1.0527,
+      "step": 1828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574577771880915e-06,
+      "loss": 1.0949,
+      "step": 1829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565233664952097e-06,
+      "loss": 1.0543,
+      "step": 1830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555889938331411e-06,
+      "loss": 1.0833,
+      "step": 1831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546546600192207e-06,
+      "loss": 1.0264,
+      "step": 1832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537203658707483e-06,
+      "loss": 1.0374,
+      "step": 1833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.527861122049901e-06,
+      "loss": 1.059,
+      "step": 1834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518518998391768e-06,
+      "loss": 1.0294,
+      "step": 1835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509177295905023e-06,
+      "loss": 1.0265,
+      "step": 1836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499836022761243e-06,
+      "loss": 1.056,
+      "step": 1837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490495187131623e-06,
+      "loss": 1.0457,
+      "step": 1838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481154797186984e-06,
+      "loss": 1.0188,
+      "step": 1839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47181486109775e-06,
+      "loss": 1.0289,
+      "step": 1840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462475387033951e-06,
+      "loss": 1.0448,
+      "step": 1841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.453136383165211e-06,
+      "loss": 0.9846,
+      "step": 1842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443797857660738e-06,
+      "loss": 0.9887,
+      "step": 1843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434459818689338e-06,
+      "loss": 1.0339,
+      "step": 1844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425122274419375e-06,
+      "loss": 1.0261,
+      "step": 1845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415785233018789e-06,
+      "loss": 0.9918,
+      "step": 1846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406448702655075e-06,
+      "loss": 1.0295,
+      "step": 1847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397112691495279e-06,
+      "loss": 1.0736,
+      "step": 1848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.387777207706006e-06,
+      "loss": 1.0636,
+      "step": 1849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378442259453384e-06,
+      "loss": 0.3769,
+      "step": 1850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369107854903083e-06,
+      "loss": 1.0397,
+      "step": 1851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359774002220289e-06,
+      "loss": 1.066,
+      "step": 1852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35044070956971e-06,
+      "loss": 1.0776,
+      "step": 1853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.341107985115569e-06,
+      "loss": 1.0368,
+      "step": 1854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331775837021588e-06,
+      "loss": 1.0222,
+      "step": 1855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322444273450976e-06,
+      "loss": 1.0558,
+      "step": 1856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31311330256644e-06,
+      "loss": 1.0286,
+      "step": 1857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303782932530172e-06,
+      "loss": 1.0309,
+      "step": 1858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.294453171503835e-06,
+      "loss": 1.0192,
+      "step": 1859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28512402764855e-06,
+      "loss": 1.0249,
+      "step": 1860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275795509124908e-06,
+      "loss": 1.0269,
+      "step": 1861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266467624092952e-06,
+      "loss": 1.087,
+      "step": 1862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257140380712171e-06,
+      "loss": 1.0277,
+      "step": 1863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247813787141491e-06,
+      "loss": 1.022,
+      "step": 1864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23848785153927e-06,
+      "loss": 1.0064,
+      "step": 1865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229162582063288e-06,
+      "loss": 1.0452,
+      "step": 1866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219837986870744e-06,
+      "loss": 1.0326,
+      "step": 1867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210514074118254e-06,
+      "loss": 1.0522,
+      "step": 1868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201190851961827e-06,
+      "loss": 1.0119,
+      "step": 1869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19186832855687e-06,
+      "loss": 1.0524,
+      "step": 1870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.182546512058184e-06,
+      "loss": 1.0105,
+      "step": 1871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173225410619945e-06,
+      "loss": 1.0599,
+      "step": 1872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163905032395711e-06,
+      "loss": 1.0142,
+      "step": 1873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154585385538402e-06,
+      "loss": 1.007,
+      "step": 1874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145266478200298e-06,
+      "loss": 1.0221,
+      "step": 1875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13594831853303e-06,
+      "loss": 1.04,
+      "step": 1876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126630914687587e-06,
+      "loss": 1.0358,
+      "step": 1877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117314274814283e-06,
+      "loss": 1.0338,
+      "step": 1878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107998407062768e-06,
+      "loss": 1.0893,
+      "step": 1879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098683319582021e-06,
+      "loss": 1.0125,
+      "step": 1880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08936902052033e-06,
+      "loss": 1.0051,
+      "step": 1881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080055518025306e-06,
+      "loss": 0.3719,
+      "step": 1882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070742820243852e-06,
+      "loss": 1.0736,
+      "step": 1883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061430935322171e-06,
+      "loss": 1.078,
+      "step": 1884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052119871405752e-06,
+      "loss": 1.0193,
+      "step": 1885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042809636639369e-06,
+      "loss": 1.0289,
+      "step": 1886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033500239167075e-06,
+      "loss": 1.0505,
+      "step": 1887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.024191687132181e-06,
+      "loss": 1.0398,
+      "step": 1888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014883988677267e-06,
+      "loss": 1.0092,
+      "step": 1889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005577151944156e-06,
+      "loss": 1.066,
+      "step": 1890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996271185073934e-06,
+      "loss": 0.9744,
+      "step": 1891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986966096206911e-06,
+      "loss": 0.9871,
+      "step": 1892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977661893482632e-06,
+      "loss": 0.9935,
+      "step": 1893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96835858503987e-06,
+      "loss": 0.946,
+      "step": 1894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959056179016612e-06,
+      "loss": 1.0166,
+      "step": 1895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949754683550065e-06,
+      "loss": 1.0124,
+      "step": 1896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940454106776625e-06,
+      "loss": 1.0688,
+      "step": 1897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9311544568319e-06,
+      "loss": 1.0472,
+      "step": 1898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921855741850673e-06,
+      "loss": 1.0359,
+      "step": 1899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912557969966912e-06,
+      "loss": 1.0529,
+      "step": 1900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903261149313775e-06,
+      "loss": 1.0347,
+      "step": 1901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89396528802357e-06,
+      "loss": 1.0089,
+      "step": 1902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.884670394227773e-06,
+      "loss": 1.037,
+      "step": 1903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875376476057008e-06,
+      "loss": 1.0509,
+      "step": 1904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866083541641056e-06,
+      "loss": 1.0574,
+      "step": 1905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85679159910883e-06,
+      "loss": 1.0183,
+      "step": 1906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84750065658838e-06,
+      "loss": 1.0172,
+      "step": 1907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838210722206876e-06,
+      "loss": 1.0391,
+      "step": 1908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828921804090605e-06,
+      "loss": 1.0209,
+      "step": 1909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819633910364977e-06,
+      "loss": 1.048,
+      "step": 1910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810347049154492e-06,
+      "loss": 1.0147,
+      "step": 1911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801061228582755e-06,
+      "loss": 0.997,
+      "step": 1912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791776456772456e-06,
+      "loss": 1.0562,
+      "step": 1913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78249274184537e-06,
+      "loss": 1.064,
+      "step": 1914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773210091922346e-06,
+      "loss": 1.0437,
+      "step": 1915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763928515123308e-06,
+      "loss": 0.9991,
+      "step": 1916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754648019567232e-06,
+      "loss": 0.9772,
+      "step": 1917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745368613372155e-06,
+      "loss": 1.0408,
+      "step": 1918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736090304655147e-06,
+      "loss": 0.3787,
+      "step": 1919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726813101532347e-06,
+      "loss": 1.0413,
+      "step": 1920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717537012118897e-06,
+      "loss": 0.9924,
+      "step": 1921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.70826204452898e-06,
+      "loss": 1.0644,
+      "step": 1922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.698988206875796e-06,
+      "loss": 1.0539,
+      "step": 1923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68971550727155e-06,
+      "loss": 0.9892,
+      "step": 1924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680443953827464e-06,
+      "loss": 1.031,
+      "step": 1925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671173554653747e-06,
+      "loss": 0.3532,
+      "step": 1926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.661904317859602e-06,
+      "loss": 1.029,
+      "step": 1927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652636251553208e-06,
+      "loss": 1.0215,
+      "step": 1928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643369363841737e-06,
+      "loss": 1.0297,
+      "step": 1929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634103662831315e-06,
+      "loss": 1.0223,
+      "step": 1930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624839156627033e-06,
+      "loss": 1.024,
+      "step": 1931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615575853332944e-06,
+      "loss": 1.0568,
+      "step": 1932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606313761052034e-06,
+      "loss": 1.0421,
+      "step": 1933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59705288788625e-06,
+      "loss": 1.0354,
+      "step": 1934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587793241936457e-06,
+      "loss": 1.0957,
+      "step": 1935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57853483130245e-06,
+      "loss": 1.0105,
+      "step": 1936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569277664082946e-06,
+      "loss": 1.0703,
+      "step": 1937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560021748375566e-06,
+      "loss": 1.0186,
+      "step": 1938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550767092276856e-06,
+      "loss": 0.3937,
+      "step": 1939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541513703882238e-06,
+      "loss": 0.9874,
+      "step": 1940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532261591286038e-06,
+      "loss": 0.3499,
+      "step": 1941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523010762581457e-06,
+      "loss": 1.021,
+      "step": 1942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513761225860586e-06,
+      "loss": 0.364,
+      "step": 1943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504512989214376e-06,
+      "loss": 1.0347,
+      "step": 1944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49526606073264e-06,
+      "loss": 1.0392,
+      "step": 1945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486020448504053e-06,
+      "loss": 1.0314,
+      "step": 1946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47677616061613e-06,
+      "loss": 0.9917,
+      "step": 1947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46753320515524e-06,
+      "loss": 1.0067,
+      "step": 1948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458291590206573e-06,
+      "loss": 1.0077,
+      "step": 1949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449051323854158e-06,
+      "loss": 0.9929,
+      "step": 1950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439812414180835e-06,
+      "loss": 1.0672,
+      "step": 1951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43057486926826e-06,
+      "loss": 0.9534,
+      "step": 1952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4213386971969e-06,
+      "loss": 1.0381,
+      "step": 1953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41210390604602e-06,
+      "loss": 1.0318,
+      "step": 1954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40287050389367e-06,
+      "loss": 1.0016,
+      "step": 1955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393638498816687e-06,
+      "loss": 1.0257,
+      "step": 1956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.384407898890694e-06,
+      "loss": 1.0248,
+      "step": 1957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.375178712190076e-06,
+      "loss": 1.021,
+      "step": 1958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36595094678799e-06,
+      "loss": 1.0084,
+      "step": 1959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356724610756342e-06,
+      "loss": 0.3683,
+      "step": 1960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347499712165785e-06,
+      "loss": 1.0024,
+      "step": 1961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338276259085729e-06,
+      "loss": 1.0433,
+      "step": 1962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329054259584306e-06,
+      "loss": 1.0138,
+      "step": 1963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319833721728385e-06,
+      "loss": 1.0723,
+      "step": 1964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310614653583546e-06,
+      "loss": 0.3844,
+      "step": 1965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301397063214094e-06,
+      "loss": 1.0191,
+      "step": 1966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292180958683035e-06,
+      "loss": 0.9881,
+      "step": 1967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.282966348052086e-06,
+      "loss": 1.0244,
+      "step": 1968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273753239381641e-06,
+      "loss": 1.0113,
+      "step": 1969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26454164073079e-06,
+      "loss": 0.3956,
+      "step": 1970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.255331560157295e-06,
+      "loss": 1.0164,
+      "step": 1971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246123005717605e-06,
+      "loss": 1.0538,
+      "step": 1972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236915985466818e-06,
+      "loss": 1.0497,
+      "step": 1973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227710507458697e-06,
+      "loss": 1.0297,
+      "step": 1974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218506579745659e-06,
+      "loss": 1.0315,
+      "step": 1975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209304210378751e-06,
+      "loss": 1.0747,
+      "step": 1976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20010340740768e-06,
+      "loss": 1.0616,
+      "step": 1977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190904178880764e-06,
+      "loss": 1.0403,
+      "step": 1978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181706532844947e-06,
+      "loss": 1.0512,
+      "step": 1979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17251047734579e-06,
+      "loss": 1.0645,
+      "step": 1980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163316020427472e-06,
+      "loss": 0.3711,
+      "step": 1981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15412317013276e-06,
+      "loss": 1.0576,
+      "step": 1982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14493193450302e-06,
+      "loss": 1.0203,
+      "step": 1983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13574232157821e-06,
+      "loss": 1.0409,
+      "step": 1984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126554339396861e-06,
+      "loss": 1.0612,
+      "step": 1985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117367995996087e-06,
+      "loss": 1.0272,
+      "step": 1986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108183299411564e-06,
+      "loss": 0.9997,
+      "step": 1987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099000257677524e-06,
+      "loss": 1.0355,
+      "step": 1988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.089818878826753e-06,
+      "loss": 1.0562,
+      "step": 1989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080639170890581e-06,
+      "loss": 1.0648,
+      "step": 1990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071461141898887e-06,
+      "loss": 1.0118,
+      "step": 1991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.062284799880068e-06,
+      "loss": 1.0169,
+      "step": 1992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053110152861054e-06,
+      "loss": 1.0459,
+      "step": 1993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043937208867287e-06,
+      "loss": 1.0479,
+      "step": 1994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034765975922718e-06,
+      "loss": 1.0641,
+      "step": 1995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02559646204981e-06,
+      "loss": 1.025,
+      "step": 1996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016428675269516e-06,
+      "loss": 1.0948,
+      "step": 1997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007262623601278e-06,
+      "loss": 1.0271,
+      "step": 1998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998098315063017e-06,
+      "loss": 1.0339,
+      "step": 1999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.988935757671144e-06,
+      "loss": 1.0381,
+      "step": 2000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979774959440518e-06,
+      "loss": 1.0463,
+      "step": 2001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.970615928384474e-06,
+      "loss": 0.9517,
+      "step": 2002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961458672514795e-06,
+      "loss": 1.0722,
+      "step": 2003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952303199841706e-06,
+      "loss": 1.0144,
+      "step": 2004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943149518373892e-06,
+      "loss": 1.033,
+      "step": 2005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.933997636118447e-06,
+      "loss": 1.0401,
+      "step": 2006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924847561080907e-06,
+      "loss": 1.0411,
+      "step": 2007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915699301265216e-06,
+      "loss": 0.3771,
+      "step": 2008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90655286467374e-06,
+      "loss": 1.0033,
+      "step": 2009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897408259307245e-06,
+      "loss": 1.0204,
+      "step": 2010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8882654931649e-06,
+      "loss": 0.981,
+      "step": 2011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87912457424426e-06,
+      "loss": 1.0432,
+      "step": 2012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869985510541258e-06,
+      "loss": 1.0531,
+      "step": 2013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860848310050223e-06,
+      "loss": 1.048,
+      "step": 2014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851712980763838e-06,
+      "loss": 0.9959,
+      "step": 2015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.842579530673156e-06,
+      "loss": 0.9828,
+      "step": 2016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833447967767577e-06,
+      "loss": 1.0232,
+      "step": 2017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824318300034867e-06,
+      "loss": 1.0397,
+      "step": 2018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81519053546112e-06,
+      "loss": 0.9745,
+      "step": 2019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806064682030773e-06,
+      "loss": 1.0433,
+      "step": 2020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796940747726587e-06,
+      "loss": 1.0043,
+      "step": 2021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787818740529645e-06,
+      "loss": 1.0193,
+      "step": 2022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778698668419338e-06,
+      "loss": 0.3986,
+      "step": 2023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769580539373384e-06,
+      "loss": 1.0771,
+      "step": 2024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760464361367782e-06,
+      "loss": 0.3609,
+      "step": 2025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75135014237683e-06,
+      "loss": 1.0069,
+      "step": 2026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.742237890373114e-06,
+      "loss": 1.0203,
+      "step": 2027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733127613327495e-06,
+      "loss": 0.9949,
+      "step": 2028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72401931920912e-06,
+      "loss": 1.0167,
+      "step": 2029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714913015985382e-06,
+      "loss": 1.026,
+      "step": 2030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705808711621947e-06,
+      "loss": 1.022,
+      "step": 2031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69670641408272e-06,
+      "loss": 1.0557,
+      "step": 2032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687606131329867e-06,
+      "loss": 1.0542,
+      "step": 2033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67850787132378e-06,
+      "loss": 1.026,
+      "step": 2034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66941164202308e-06,
+      "loss": 1.0296,
+      "step": 2035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660317451384622e-06,
+      "loss": 1.0289,
+      "step": 2036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651225307363463e-06,
+      "loss": 0.9938,
+      "step": 2037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642135217912887e-06,
+      "loss": 0.3572,
+      "step": 2038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633047190984369e-06,
+      "loss": 1.0132,
+      "step": 2039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623961234527581e-06,
+      "loss": 0.9545,
+      "step": 2040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614877356490387e-06,
+      "loss": 1.0279,
+      "step": 2041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.605795564818831e-06,
+      "loss": 0.3902,
+      "step": 2042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596715867457137e-06,
+      "loss": 1.0141,
+      "step": 2043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587638272347684e-06,
+      "loss": 0.3596,
+      "step": 2044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.578562787431028e-06,
+      "loss": 1.0181,
+      "step": 2045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5694894206458705e-06,
+      "loss": 1.0346,
+      "step": 2046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560418179929057e-06,
+      "loss": 1.0014,
+      "step": 2047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551349073215584e-06,
+      "loss": 1.0383,
+      "step": 2048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.542282108438572e-06,
+      "loss": 0.9708,
+      "step": 2049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53321729352927e-06,
+      "loss": 1.0613,
+      "step": 2050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524154636417048e-06,
+      "loss": 0.3773,
+      "step": 2051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515094145029391e-06,
+      "loss": 1.0344,
+      "step": 2052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506035827291882e-06,
+      "loss": 0.3499,
+      "step": 2053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49697969112821e-06,
+      "loss": 1.0584,
+      "step": 2054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487925744460152e-06,
+      "loss": 0.993,
+      "step": 2055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478873995207565e-06,
+      "loss": 1.0404,
+      "step": 2056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469824451288398e-06,
+      "loss": 1.0428,
+      "step": 2057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460777120618658e-06,
+      "loss": 1.0472,
+      "step": 2058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.451732011112419e-06,
+      "loss": 1.043,
+      "step": 2059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4426891306818176e-06,
+      "loss": 0.9976,
+      "step": 2060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.433648487237028e-06,
+      "loss": 1.0422,
+      "step": 2061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.424610088686284e-06,
+      "loss": 1.0463,
+      "step": 2062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4155739429358455e-06,
+      "loss": 1.0044,
+      "step": 2063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406540057890001e-06,
+      "loss": 1.0411,
+      "step": 2064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397508441451065e-06,
+      "loss": 1.0067,
+      "step": 2065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388479101519368e-06,
+      "loss": 1.0468,
+      "step": 2066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379452045993249e-06,
+      "loss": 0.9905,
+      "step": 2067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3704272827690435e-06,
+      "loss": 0.3468,
+      "step": 2068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361404819741091e-06,
+      "loss": 1.0164,
+      "step": 2069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35238466480171e-06,
+      "loss": 1.0582,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.343366825841205e-06,
+      "loss": 1.014,
+      "step": 2071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334351310747856e-06,
+      "loss": 0.9904,
+      "step": 2072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325338127407904e-06,
+      "loss": 1.0248,
+      "step": 2073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316327283705558e-06,
+      "loss": 1.0481,
+      "step": 2074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307318787522968e-06,
+      "loss": 1.0021,
+      "step": 2075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29831264674025e-06,
+      "loss": 1.0128,
+      "step": 2076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289308869235441e-06,
+      "loss": 1.0572,
+      "step": 2077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280307462884524e-06,
+      "loss": 1.0193,
+      "step": 2078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271308435561399e-06,
+      "loss": 0.9755,
+      "step": 2079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2623117951378856e-06,
+      "loss": 0.3691,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253317549483724e-06,
+      "loss": 1.02,
+      "step": 2081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244325706466551e-06,
+      "loss": 1.0122,
+      "step": 2082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235336273951905e-06,
+      "loss": 1.0308,
+      "step": 2083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226349259803209e-06,
+      "loss": 1.0122,
+      "step": 2084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217364671881787e-06,
+      "loss": 1.0179,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208382518046824e-06,
+      "loss": 0.3891,
+      "step": 2086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199402806155384e-06,
+      "loss": 1.0182,
+      "step": 2087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.190425544062395e-06,
+      "loss": 1.0243,
+      "step": 2088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181450739620634e-06,
+      "loss": 1.0098,
+      "step": 2089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1724784006807425e-06,
+      "loss": 1.026,
+      "step": 2090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1635085350911935e-06,
+      "loss": 1.0588,
+      "step": 2091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1545411506983e-06,
+      "loss": 0.3602,
+      "step": 2092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145576255346205e-06,
+      "loss": 1.0243,
+      "step": 2093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136613856876872e-06,
+      "loss": 1.0072,
+      "step": 2094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127653963130086e-06,
+      "loss": 1.064,
+      "step": 2095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.118696581943436e-06,
+      "loss": 0.999,
+      "step": 2096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109741721152317e-06,
+      "loss": 1.0189,
+      "step": 2097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1007893885899134e-06,
+      "loss": 1.0413,
+      "step": 2098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0918395920871996e-06,
+      "loss": 0.9786,
+      "step": 2099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08289233947294e-06,
+      "loss": 1.0224,
+      "step": 2100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.073947638573664e-06,
+      "loss": 1.0236,
+      "step": 2101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065005497213671e-06,
+      "loss": 1.0331,
+      "step": 2102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056065923215023e-06,
+      "loss": 1.0652,
+      "step": 2103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047128924397537e-06,
+      "loss": 0.348,
+      "step": 2104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038194508578777e-06,
+      "loss": 1.0347,
+      "step": 2105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029262683574049e-06,
+      "loss": 1.0218,
+      "step": 2106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0203334571963855e-06,
+      "loss": 1.0366,
+      "step": 2107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011406837256549e-06,
+      "loss": 1.0014,
+      "step": 2108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0024828315630325e-06,
+      "loss": 1.0549,
+      "step": 2109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993561447922028e-06,
+      "loss": 1.055,
+      "step": 2110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9846426941374405e-06,
+      "loss": 1.0074,
+      "step": 2111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9757265780108755e-06,
+      "loss": 0.9576,
+      "step": 2112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966813107341622e-06,
+      "loss": 1.0163,
+      "step": 2113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957902289926674e-06,
+      "loss": 1.0081,
+      "step": 2114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948994133560685e-06,
+      "loss": 1.0004,
+      "step": 2115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940088646035992e-06,
+      "loss": 0.9921,
+      "step": 2116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9311858351425906e-06,
+      "loss": 0.9963,
+      "step": 2117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922285708668137e-06,
+      "loss": 0.982,
+      "step": 2118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91338827439795e-06,
+      "loss": 1.0025,
+      "step": 2119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904493540114974e-06,
+      "loss": 1.0121,
+      "step": 2120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895601513599808e-06,
+      "loss": 1.0181,
+      "step": 2121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886712202630671e-06,
+      "loss": 1.0284,
+      "step": 2122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8778256149834175e-06,
+      "loss": 1.0105,
+      "step": 2123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868941758431513e-06,
+      "loss": 1.0369,
+      "step": 2124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.860060640746034e-06,
+      "loss": 1.0075,
+      "step": 2125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85118226969566e-06,
+      "loss": 1.0161,
+      "step": 2126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842306653046669e-06,
+      "loss": 1.0091,
+      "step": 2127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833433798562939e-06,
+      "loss": 1.0521,
+      "step": 2128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824563714005914e-06,
+      "loss": 1.0009,
+      "step": 2129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.815696407134631e-06,
+      "loss": 1.0669,
+      "step": 2130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.80683188570569e-06,
+      "loss": 1.0321,
+      "step": 2131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.79797015747325e-06,
+      "loss": 1.0116,
+      "step": 2132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789111230189039e-06,
+      "loss": 0.3778,
+      "step": 2133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780255111602326e-06,
+      "loss": 0.9734,
+      "step": 2134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771401809459925e-06,
+      "loss": 0.9941,
+      "step": 2135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76255133150618e-06,
+      "loss": 1.045,
+      "step": 2136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75370368548298e-06,
+      "loss": 1.0676,
+      "step": 2137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744858879129723e-06,
+      "loss": 1.0358,
+      "step": 2138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73601692018333e-06,
+      "loss": 1.0312,
+      "step": 2139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.727177816378228e-06,
+      "loss": 1.0255,
+      "step": 2140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7183415754463434e-06,
+      "loss": 0.9842,
+      "step": 2141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70950820511711e-06,
+      "loss": 0.9971,
+      "step": 2142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.700677713117439e-06,
+      "loss": 0.9857,
+      "step": 2143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691850107171728e-06,
+      "loss": 0.9776,
+      "step": 2144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6830253950018475e-06,
+      "loss": 1.0246,
+      "step": 2145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674203584327141e-06,
+      "loss": 1.0501,
+      "step": 2146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.665384682864413e-06,
+      "loss": 1.0544,
+      "step": 2147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656568698327923e-06,
+      "loss": 1.0167,
+      "step": 2148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6477556384293774e-06,
+      "loss": 0.3651,
+      "step": 2149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638945510877924e-06,
+      "loss": 0.3788,
+      "step": 2150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630138323380142e-06,
+      "loss": 0.9655,
+      "step": 2151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621334083640052e-06,
+      "loss": 1.0376,
+      "step": 2152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6125327993590825e-06,
+      "loss": 0.991,
+      "step": 2153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6037344782360815e-06,
+      "loss": 1.0209,
+      "step": 2154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594939127967306e-06,
+      "loss": 1.0741,
+      "step": 2155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586146756246415e-06,
+      "loss": 1.0212,
+      "step": 2156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577357370764459e-06,
+      "loss": 1.0679,
+      "step": 2157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568570979209879e-06,
+      "loss": 1.0142,
+      "step": 2158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559787589268496e-06,
+      "loss": 1.032,
+      "step": 2159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551007208623498e-06,
+      "loss": 0.3526,
+      "step": 2160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542229844955459e-06,
+      "loss": 1.0113,
+      "step": 2161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.533455505942298e-06,
+      "loss": 1.0315,
+      "step": 2162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52468419925929e-06,
+      "loss": 1.0701,
+      "step": 2163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515915932579067e-06,
+      "loss": 1.0426,
+      "step": 2164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.507150713571586e-06,
+      "loss": 0.9834,
+      "step": 2165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.498388549904158e-06,
+      "loss": 1.0273,
+      "step": 2166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489629449241406e-06,
+      "loss": 1.0219,
+      "step": 2167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480873419245277e-06,
+      "loss": 1.0583,
+      "step": 2168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472120467575036e-06,
+      "loss": 0.9931,
+      "step": 2169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463370601887247e-06,
+      "loss": 1.0233,
+      "step": 2170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454623829835787e-06,
+      "loss": 0.3684,
+      "step": 2171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445880159071817e-06,
+      "loss": 1.0279,
+      "step": 2172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437139597243789e-06,
+      "loss": 0.9947,
+      "step": 2173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428402151997433e-06,
+      "loss": 1.0461,
+      "step": 2174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4196678309757575e-06,
+      "loss": 1.0381,
+      "step": 2175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410936641819034e-06,
+      "loss": 1.0188,
+      "step": 2176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402208592164797e-06,
+      "loss": 0.3603,
+      "step": 2177
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393483689647831e-06,
+      "loss": 1.0229,
+      "step": 2178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384761941900165e-06,
+      "loss": 1.0429,
+      "step": 2179
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376043356551083e-06,
+      "loss": 0.9792,
+      "step": 2180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367327941227086e-06,
+      "loss": 1.0382,
+      "step": 2181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358615703551916e-06,
+      "loss": 1.0065,
+      "step": 2182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349906651146523e-06,
+      "loss": 0.3569,
+      "step": 2183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341200791629072e-06,
+      "loss": 1.0387,
+      "step": 2184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332498132614948e-06,
+      "loss": 1.0062,
+      "step": 2185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3237986817167255e-06,
+      "loss": 1.0231,
+      "step": 2186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315102446544173e-06,
+      "loss": 0.9778,
+      "step": 2187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306409434704244e-06,
+      "loss": 0.9896,
+      "step": 2188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297719653801083e-06,
+      "loss": 1.0702,
+      "step": 2189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.289033111435998e-06,
+      "loss": 1.0107,
+      "step": 2190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28034981520747e-06,
+      "loss": 1.0465,
+      "step": 2191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271669772711137e-06,
+      "loss": 1.032,
+      "step": 2192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262992991539788e-06,
+      "loss": 1.0098,
+      "step": 2193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25431947928337e-06,
+      "loss": 1.0303,
+      "step": 2194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245649243528962e-06,
+      "loss": 1.016,
+      "step": 2195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236982291860776e-06,
+      "loss": 1.0296,
+      "step": 2196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2283186318601514e-06,
+      "loss": 0.9684,
+      "step": 2197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219658271105555e-06,
+      "loss": 1.0123,
+      "step": 2198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.211001217172564e-06,
+      "loss": 1.0061,
+      "step": 2199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.202347477633862e-06,
+      "loss": 0.9913,
+      "step": 2200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193697060059233e-06,
+      "loss": 0.9417,
+      "step": 2201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1850499720155546e-06,
+      "loss": 1.0206,
+      "step": 2202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17640622106679e-06,
+      "loss": 1.0053,
+      "step": 2203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167765814773994e-06,
+      "loss": 1.0495,
+      "step": 2204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1591287606952855e-06,
+      "loss": 1.0329,
+      "step": 2205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150495066385848e-06,
+      "loss": 0.3512,
+      "step": 2206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141864739397939e-06,
+      "loss": 1.0297,
+      "step": 2207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13323778728086e-06,
+      "loss": 0.9974,
+      "step": 2208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124614217580966e-06,
+      "loss": 0.9954,
+      "step": 2209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115994037841649e-06,
+      "loss": 1.0022,
+      "step": 2210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.107377255603338e-06,
+      "loss": 0.9921,
+      "step": 2211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0987638784034846e-06,
+      "loss": 1.0327,
+      "step": 2212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.090153913776576e-06,
+      "loss": 1.0313,
+      "step": 2213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081547369254101e-06,
+      "loss": 1.004,
+      "step": 2214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072944252364556e-06,
+      "loss": 1.0384,
+      "step": 2215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064344570633454e-06,
+      "loss": 1.0028,
+      "step": 2216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055748331583281e-06,
+      "loss": 1.0099,
+      "step": 2217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.047155542733534e-06,
+      "loss": 0.9691,
+      "step": 2218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038566211600678e-06,
+      "loss": 1.0745,
+      "step": 2219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029980345698158e-06,
+      "loss": 0.9965,
+      "step": 2220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0213979525363835e-06,
+      "loss": 0.3834,
+      "step": 2221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012819039622728e-06,
+      "loss": 0.3672,
+      "step": 2222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0042436144615305e-06,
+      "loss": 1.005,
+      "step": 2223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995671684554063e-06,
+      "loss": 0.9982,
+      "step": 2224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987103257398552e-06,
+      "loss": 1.0398,
+      "step": 2225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978538340490152e-06,
+      "loss": 0.3547,
+      "step": 2226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.969976941320957e-06,
+      "loss": 0.9801,
+      "step": 2227
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961419067379974e-06,
+      "loss": 1.0325,
+      "step": 2228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952864726153131e-06,
+      "loss": 0.9945,
+      "step": 2229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944313925123265e-06,
+      "loss": 1.0226,
+      "step": 2230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93576667177011e-06,
+      "loss": 0.9879,
+      "step": 2231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927222973570312e-06,
+      "loss": 1.0106,
+      "step": 2232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.918682837997394e-06,
+      "loss": 1.0363,
+      "step": 2233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.910146272521766e-06,
+      "loss": 1.0524,
+      "step": 2234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9016132846107166e-06,
+      "loss": 1.0104,
+      "step": 2235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893083881728401e-06,
+      "loss": 1.0584,
+      "step": 2236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884558071335847e-06,
+      "loss": 0.9943,
+      "step": 2237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8760358608909315e-06,
+      "loss": 1.0167,
+      "step": 2238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.867517257848385e-06,
+      "loss": 1.0053,
+      "step": 2239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8590022696597826e-06,
+      "loss": 1.0051,
+      "step": 2240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850490903773536e-06,
+      "loss": 0.9888,
+      "step": 2241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8419831676348924e-06,
+      "loss": 1.0196,
+      "step": 2242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833479068685923e-06,
+      "loss": 1.0386,
+      "step": 2243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82497861436551e-06,
+      "loss": 0.9948,
+      "step": 2244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816481812109354e-06,
+      "loss": 1.0471,
+      "step": 2245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807988669349964e-06,
+      "loss": 1.0436,
+      "step": 2246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799499193516642e-06,
+      "loss": 0.9757,
+      "step": 2247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7910133920354886e-06,
+      "loss": 1.0122,
+      "step": 2248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.782531272329377e-06,
+      "loss": 1.0132,
+      "step": 2249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7740528418179745e-06,
+      "loss": 1.0114,
+      "step": 2250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765578107917715e-06,
+      "loss": 1.0297,
+      "step": 2251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7571070780418036e-06,
+      "loss": 1.0519,
+      "step": 2252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748639759600194e-06,
+      "loss": 1.0807,
+      "step": 2253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740176159999602e-06,
+      "loss": 0.9886,
+      "step": 2254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731716286643493e-06,
+      "loss": 0.3606,
+      "step": 2255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723260146932066e-06,
+      "loss": 1.0514,
+      "step": 2256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714807748262261e-06,
+      "loss": 1.0068,
+      "step": 2257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.706359098027734e-06,
+      "loss": 0.9649,
+      "step": 2258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697914203618875e-06,
+      "loss": 1.0329,
+      "step": 2259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689473072422781e-06,
+      "loss": 0.9682,
+      "step": 2260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.681035711823264e-06,
+      "loss": 1.0434,
+      "step": 2261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672602129200826e-06,
+      "loss": 1.0111,
+      "step": 2262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6641723319326735e-06,
+      "loss": 1.0083,
+      "step": 2263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655746327392701e-06,
+      "loss": 1.0101,
+      "step": 2264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647324122951484e-06,
+      "loss": 1.0514,
+      "step": 2265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638905725976277e-06,
+      "loss": 0.3906,
+      "step": 2266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630491143830994e-06,
+      "loss": 1.0411,
+      "step": 2267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622080383876224e-06,
+      "loss": 1.0303,
+      "step": 2268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.613673453469198e-06,
+      "loss": 1.0197,
+      "step": 2269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.605270359963823e-06,
+      "loss": 1.037,
+      "step": 2270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596871110710618e-06,
+      "loss": 1.0184,
+      "step": 2271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5884757130567625e-06,
+      "loss": 1.0056,
+      "step": 2272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.580084174346058e-06,
+      "loss": 1.0269,
+      "step": 2273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571696501918924e-06,
+      "loss": 1.0018,
+      "step": 2274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563312703112419e-06,
+      "loss": 0.9524,
+      "step": 2275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554932785260187e-06,
+      "loss": 0.9973,
+      "step": 2276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546556755692498e-06,
+      "loss": 1.0332,
+      "step": 2277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.538184621736201e-06,
+      "loss": 1.0479,
+      "step": 2278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529816390714764e-06,
+      "loss": 0.3697,
+      "step": 2279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521452069948212e-06,
+      "loss": 1.0162,
+      "step": 2280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51309166675317e-06,
+      "loss": 0.9681,
+      "step": 2281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.50473518844283e-06,
+      "loss": 1.0228,
+      "step": 2282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.49638264232694e-06,
+      "loss": 1.0289,
+      "step": 2283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488034035711835e-06,
+      "loss": 0.9995,
+      "step": 2284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4796893759003735e-06,
+      "loss": 1.0596,
+      "step": 2285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471348670191984e-06,
+      "loss": 1.0299,
+      "step": 2286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463011925882619e-06,
+      "loss": 1.0023,
+      "step": 2287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454679150264779e-06,
+      "loss": 1.0135,
+      "step": 2288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446350350627489e-06,
+      "loss": 1.0461,
+      "step": 2289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4380255342562935e-06,
+      "loss": 1.031,
+      "step": 2290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.429704708433259e-06,
+      "loss": 1.0147,
+      "step": 2291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42138788043695e-06,
+      "loss": 1.0125,
+      "step": 2292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413075057542443e-06,
+      "loss": 1.0291,
+      "step": 2293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404766247021309e-06,
+      "loss": 1.0038,
+      "step": 2294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396461456141613e-06,
+      "loss": 1.0532,
+      "step": 2295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3881606921678934e-06,
+      "loss": 0.9756,
+      "step": 2296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3798639623611734e-06,
+      "loss": 0.963,
+      "step": 2297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3715712739789505e-06,
+      "loss": 0.362,
+      "step": 2298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36328263427518e-06,
+      "loss": 0.9971,
+      "step": 2299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.354998050500283e-06,
+      "loss": 0.9976,
+      "step": 2300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346717529901121e-06,
+      "loss": 0.99,
+      "step": 2301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338441079721013e-06,
+      "loss": 1.0105,
+      "step": 2302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.330168707199711e-06,
+      "loss": 1.0332,
+      "step": 2303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.321900419573408e-06,
+      "loss": 1.0155,
+      "step": 2304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3136362240747095e-06,
+      "loss": 0.998,
+      "step": 2305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305376127932653e-06,
+      "loss": 1.0051,
+      "step": 2306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297120138372689e-06,
+      "loss": 1.0006,
+      "step": 2307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288868262616672e-06,
+      "loss": 0.9923,
+      "step": 2308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280620507882864e-06,
+      "loss": 1.0466,
+      "step": 2309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.27237688138591e-06,
+      "loss": 1.0262,
+      "step": 2310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264137390336861e-06,
+      "loss": 1.0094,
+      "step": 2311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255902041943127e-06,
+      "loss": 0.9862,
+      "step": 2312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2476708434085265e-06,
+      "loss": 0.9773,
+      "step": 2313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239443801933217e-06,
+      "loss": 1.0116,
+      "step": 2314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231220924713738e-06,
+      "loss": 0.9784,
+      "step": 2315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.223002218942979e-06,
+      "loss": 1.0221,
+      "step": 2316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2147876918101844e-06,
+      "loss": 1.0469,
+      "step": 2317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206577350500945e-06,
+      "loss": 1.0116,
+      "step": 2318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198371202197179e-06,
+      "loss": 1.0263,
+      "step": 2319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190169254077151e-06,
+      "loss": 1.0461,
+      "step": 2320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181971513315435e-06,
+      "loss": 0.3224,
+      "step": 2321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1737779870829485e-06,
+      "loss": 0.3542,
+      "step": 2322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165588682546898e-06,
+      "loss": 0.9918,
+      "step": 2323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157403606870812e-06,
+      "loss": 1.0036,
+      "step": 2324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149222767214519e-06,
+      "loss": 1.0127,
+      "step": 2325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.141046170734125e-06,
+      "loss": 1.0104,
+      "step": 2326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132873824582056e-06,
+      "loss": 1.0226,
+      "step": 2327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12470573590699e-06,
+      "loss": 1.0123,
+      "step": 2328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116541911853898e-06,
+      "loss": 1.0198,
+      "step": 2329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1083823595640055e-06,
+      "loss": 1.0126,
+      "step": 2330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1002270861748285e-06,
+      "loss": 1.0315,
+      "step": 2331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092076098820107e-06,
+      "loss": 1.0046,
+      "step": 2332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083929404629855e-06,
+      "loss": 0.9956,
+      "step": 2333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075787010730324e-06,
+      "loss": 1.0205,
+      "step": 2334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067648924243994e-06,
+      "loss": 0.9775,
+      "step": 2335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059515152289603e-06,
+      "loss": 1.0375,
+      "step": 2336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051385701982083e-06,
+      "loss": 1.0215,
+      "step": 2337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.043260580432614e-06,
+      "loss": 1.0213,
+      "step": 2338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.035139794748566e-06,
+      "loss": 1.0308,
+      "step": 2339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02702335203353e-06,
+      "loss": 1.0362,
+      "step": 2340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0189112593872954e-06,
+      "loss": 0.3729,
+      "step": 2341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010803523905848e-06,
+      "loss": 1.0432,
+      "step": 2342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002700152681361e-06,
+      "loss": 1.0448,
+      "step": 2343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994601152802185e-06,
+      "loss": 1.0101,
+      "step": 2344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986506531352849e-06,
+      "loss": 1.0237,
+      "step": 2345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97841629541406e-06,
+      "loss": 1.005,
+      "step": 2346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9703304520626826e-06,
+      "loss": 0.9988,
+      "step": 2347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9622490083717335e-06,
+      "loss": 1.0153,
+      "step": 2348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954171971410388e-06,
+      "loss": 1.046,
+      "step": 2349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946099348243969e-06,
+      "loss": 1.0016,
+      "step": 2350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9380311459339315e-06,
+      "loss": 1.0382,
+      "step": 2351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929967371537871e-06,
+      "loss": 1.0344,
+      "step": 2352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921908032109497e-06,
+      "loss": 0.9959,
+      "step": 2353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9138531346986515e-06,
+      "loss": 0.9878,
+      "step": 2354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9058026863512894e-06,
+      "loss": 0.9976,
+      "step": 2355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897756694109474e-06,
+      "loss": 1.0055,
+      "step": 2356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889715165011359e-06,
+      "loss": 0.9776,
+      "step": 2357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881678106091209e-06,
+      "loss": 1.0316,
+      "step": 2358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873645524379372e-06,
+      "loss": 0.9753,
+      "step": 2359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865617426902281e-06,
+      "loss": 1.0118,
+      "step": 2360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857593820682448e-06,
+      "loss": 1.0687,
+      "step": 2361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8495747127384474e-06,
+      "loss": 1.0221,
+      "step": 2362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841560110084932e-06,
+      "loss": 1.0181,
+      "step": 2363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833550019732596e-06,
+      "loss": 0.9873,
+      "step": 2364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825544448688212e-06,
+      "loss": 1.0323,
+      "step": 2365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817543403954574e-06,
+      "loss": 0.9907,
+      "step": 2366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809546892530534e-06,
+      "loss": 1.0001,
+      "step": 2367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801554921410966e-06,
+      "loss": 1.0194,
+      "step": 2368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793567497586783e-06,
+      "loss": 1.014,
+      "step": 2369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785584628044918e-06,
+      "loss": 1.0336,
+      "step": 2370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777606319768311e-06,
+      "loss": 1.0364,
+      "step": 2371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7696325797359265e-06,
+      "loss": 0.354,
+      "step": 2372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.761663414922713e-06,
+      "loss": 1.047,
+      "step": 2373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.753698832299645e-06,
+      "loss": 1.0494,
+      "step": 2374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7457388388336615e-06,
+      "loss": 1.0266,
+      "step": 2375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737783441487702e-06,
+      "loss": 0.9915,
+      "step": 2376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729832647220688e-06,
+      "loss": 1.0498,
+      "step": 2377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721886462987496e-06,
+      "loss": 0.9546,
+      "step": 2378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713944895738998e-06,
+      "loss": 1.0274,
+      "step": 2379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7060079524220015e-06,
+      "loss": 1.0392,
+      "step": 2380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698075639979286e-06,
+      "loss": 1.0278,
+      "step": 2381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690147965349563e-06,
+      "loss": 1.0159,
+      "step": 2382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682224935467513e-06,
+      "loss": 0.3202,
+      "step": 2383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674306557263729e-06,
+      "loss": 0.976,
+      "step": 2384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666392837664747e-06,
+      "loss": 0.3733,
+      "step": 2385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6584837835930295e-06,
+      "loss": 0.9865,
+      "step": 2386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6505794019669424e-06,
+      "loss": 0.969,
+      "step": 2387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642679699700792e-06,
+      "loss": 1.0365,
+      "step": 2388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6347846837047624e-06,
+      "loss": 0.999,
+      "step": 2389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62689436088496e-06,
+      "loss": 0.9286,
+      "step": 2390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61900873814337e-06,
+      "loss": 1.0183,
+      "step": 2391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611127822377877e-06,
+      "loss": 0.9642,
+      "step": 2392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6032516204822445e-06,
+      "loss": 1.0063,
+      "step": 2393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5953801393461105e-06,
+      "loss": 1.026,
+      "step": 2394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587513385854994e-06,
+      "loss": 1.0228,
+      "step": 2395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57965136689026e-06,
+      "loss": 0.996,
+      "step": 2396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571794089329146e-06,
+      "loss": 0.991,
+      "step": 2397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563941560044739e-06,
+      "loss": 1.0258,
+      "step": 2398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556093785905976e-06,
+      "loss": 1.0181,
+      "step": 2399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548250773777623e-06,
+      "loss": 0.9962,
+      "step": 2400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540412530520291e-06,
+      "loss": 1.0327,
+      "step": 2401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532579062990415e-06,
+      "loss": 0.9827,
+      "step": 2402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5247503780402566e-06,
+      "loss": 1.0223,
+      "step": 2403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516926482517893e-06,
+      "loss": 0.9725,
+      "step": 2404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5091073832672016e-06,
+      "loss": 1.0416,
+      "step": 2405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501293087127878e-06,
+      "loss": 0.999,
+      "step": 2406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.49348360093541e-06,
+      "loss": 1.0486,
+      "step": 2407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.485678931521084e-06,
+      "loss": 1.0523,
+      "step": 2408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477879085711957e-06,
+      "loss": 0.9824,
+      "step": 2409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470084070330886e-06,
+      "loss": 0.981,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4622938921964895e-06,
+      "loss": 1.0071,
+      "step": 2411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454508558123163e-06,
+      "loss": 1.0168,
+      "step": 2412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446728074921062e-06,
+      "loss": 0.9844,
+      "step": 2413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43895244939609e-06,
+      "loss": 1.0009,
+      "step": 2414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431181688349919e-06,
+      "loss": 1.0708,
+      "step": 2415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423415798579942e-06,
+      "loss": 1.0558,
+      "step": 2416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415654786879321e-06,
+      "loss": 1.0017,
+      "step": 2417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407898660036923e-06,
+      "loss": 0.9858,
+      "step": 2418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400147424837359e-06,
+      "loss": 1.0325,
+      "step": 2419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392401088060953e-06,
+      "loss": 1.034,
+      "step": 2420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384659656483747e-06,
+      "loss": 1.0385,
+      "step": 2421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376923136877498e-06,
+      "loss": 0.9759,
+      "step": 2422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369191536009651e-06,
+      "loss": 1.0271,
+      "step": 2423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361464860643366e-06,
+      "loss": 0.9928,
+      "step": 2424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353743117537477e-06,
+      "loss": 0.9852,
+      "step": 2425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346026313446523e-06,
+      "loss": 1.0363,
+      "step": 2426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338314455120706e-06,
+      "loss": 1.0242,
+      "step": 2427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3306075493059075e-06,
+      "loss": 1.0137,
+      "step": 2428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322905602743684e-06,
+      "loss": 0.9845,
+      "step": 2429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315208622171238e-06,
+      "loss": 1.0218,
+      "step": 2430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3075166143214475e-06,
+      "loss": 1.0299,
+      "step": 2431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299829585922822e-06,
+      "loss": 1.0046,
+      "step": 2432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292147543699532e-06,
+      "loss": 0.9672,
+      "step": 2433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28447049437137e-06,
+      "loss": 0.9818,
+      "step": 2434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276798444653772e-06,
+      "loss": 0.9677,
+      "step": 2435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2691314012578015e-06,
+      "loss": 1.0095,
+      "step": 2436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2614693708901345e-06,
+      "loss": 0.9666,
+      "step": 2437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253812360253072e-06,
+      "loss": 0.9917,
+      "step": 2438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246160376044509e-06,
+      "loss": 1.0101,
+      "step": 2439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238513424957966e-06,
+      "loss": 0.9879,
+      "step": 2440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230871513682538e-06,
+      "loss": 1.0176,
+      "step": 2441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.223234648902929e-06,
+      "loss": 1.0307,
+      "step": 2442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2156028372994125e-06,
+      "loss": 0.9964,
+      "step": 2443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207976085547857e-06,
+      "loss": 0.9901,
+      "step": 2444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200354400319695e-06,
+      "loss": 0.9796,
+      "step": 2445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192737788281933e-06,
+      "loss": 1.0436,
+      "step": 2446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18512625609714e-06,
+      "loss": 0.9984,
+      "step": 2447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177519810423431e-06,
+      "loss": 1.0498,
+      "step": 2448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169918457914482e-06,
+      "loss": 0.3664,
+      "step": 2449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1623222052195144e-06,
+      "loss": 1.0181,
+      "step": 2450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154731058983287e-06,
+      "loss": 0.365,
+      "step": 2451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147145025846082e-06,
+      "loss": 1.0597,
+      "step": 2452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139564112443721e-06,
+      "loss": 1.0033,
+      "step": 2453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131988325407544e-06,
+      "loss": 1.0365,
+      "step": 2454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1244176713644055e-06,
+      "loss": 0.9897,
+      "step": 2455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1168521569366745e-06,
+      "loss": 1.0383,
+      "step": 2456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109291788742213e-06,
+      "loss": 0.9826,
+      "step": 2457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101736573394389e-06,
+      "loss": 0.9893,
+      "step": 2458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094186517502067e-06,
+      "loss": 1.0036,
+      "step": 2459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.086641627669594e-06,
+      "loss": 0.9876,
+      "step": 2460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079101910496793e-06,
+      "loss": 1.014,
+      "step": 2461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07156737257897e-06,
+      "loss": 1.0394,
+      "step": 2462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0640380205068965e-06,
+      "loss": 0.9373,
+      "step": 2463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056513860866809e-06,
+      "loss": 1.0065,
+      "step": 2464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0489949002404085e-06,
+      "loss": 0.9544,
+      "step": 2465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041481145204831e-06,
+      "loss": 1.0368,
+      "step": 2466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033972602332677e-06,
+      "loss": 0.3608,
+      "step": 2467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02646927819197e-06,
+      "loss": 0.3683,
+      "step": 2468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018971179346194e-06,
+      "loss": 0.3642,
+      "step": 2469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011478312354235e-06,
+      "loss": 0.9982,
+      "step": 2470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003990683770417e-06,
+      "loss": 1.031,
+      "step": 2471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996508300144481e-06,
+      "loss": 1.0303,
+      "step": 2472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989031168021576e-06,
+      "loss": 1.0529,
+      "step": 2473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9815592939422655e-06,
+      "loss": 0.9824,
+      "step": 2474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9740926844424976e-06,
+      "loss": 0.9318,
+      "step": 2475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966631346053635e-06,
+      "loss": 0.3686,
+      "step": 2476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.959175285302408e-06,
+      "loss": 0.998,
+      "step": 2477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.951724508710955e-06,
+      "loss": 1.0113,
+      "step": 2478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944279022796768e-06,
+      "loss": 0.9996,
+      "step": 2479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936838834072729e-06,
+      "loss": 1.013,
+      "step": 2480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9294039490470805e-06,
+      "loss": 1.0278,
+      "step": 2481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921974374223414e-06,
+      "loss": 1.0505,
+      "step": 2482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914550116100702e-06,
+      "loss": 0.9908,
+      "step": 2483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907131181173239e-06,
+      "loss": 1.0103,
+      "step": 2484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8997175759306826e-06,
+      "loss": 1.0483,
+      "step": 2485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892309306858012e-06,
+      "loss": 1.0306,
+      "step": 2486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884906380435553e-06,
+      "loss": 0.3725,
+      "step": 2487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877508803138952e-06,
+      "loss": 1.0091,
+      "step": 2488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870116581439175e-06,
+      "loss": 0.9811,
+      "step": 2489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862729721802509e-06,
+      "loss": 0.9809,
+      "step": 2490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855348230690535e-06,
+      "loss": 0.9595,
+      "step": 2491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8479721145601654e-06,
+      "loss": 1.0158,
+      "step": 2492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8406013798635824e-06,
+      "loss": 0.9696,
+      "step": 2493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83323603304828e-06,
+      "loss": 0.9895,
+      "step": 2494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825876080557026e-06,
+      "loss": 0.9755,
+      "step": 2495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81852152882788e-06,
+      "loss": 0.3969,
+      "step": 2496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811172384294172e-06,
+      "loss": 0.9956,
+      "step": 2497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803828653384504e-06,
+      "loss": 1.0142,
+      "step": 2498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796490342522745e-06,
+      "loss": 0.9534,
+      "step": 2499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7891574581280135e-06,
+      "loss": 0.9848,
+      "step": 2500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7818300066146894e-06,
+      "loss": 0.9879,
+      "step": 2501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7745079943924003e-06,
+      "loss": 0.9524,
+      "step": 2502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767191427866015e-06,
+      "loss": 0.9952,
+      "step": 2503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7598803134356323e-06,
+      "loss": 1.0122,
+      "step": 2504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7525746574965906e-06,
+      "loss": 1.0397,
+      "step": 2505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745274466439448e-06,
+      "loss": 0.9701,
+      "step": 2506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737979746649987e-06,
+      "loss": 0.9695,
+      "step": 2507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7306905045092023e-06,
+      "loss": 1.0073,
+      "step": 2508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.72340674639329e-06,
+      "loss": 0.9967,
+      "step": 2509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7161284786736598e-06,
+      "loss": 1.0151,
+      "step": 2510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.708855707716913e-06,
+      "loss": 0.9867,
+      "step": 2511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7015884398848466e-06,
+      "loss": 0.3525,
+      "step": 2512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694326681534436e-06,
+      "loss": 0.9596,
+      "step": 2513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6870704390178447e-06,
+      "loss": 1.0141,
+      "step": 2514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679819718682409e-06,
+      "loss": 1.0016,
+      "step": 2515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6725745268706337e-06,
+      "loss": 0.9959,
+      "step": 2516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665334869920193e-06,
+      "loss": 1.0013,
+      "step": 2517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658100754163907e-06,
+      "loss": 0.9575,
+      "step": 2518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6508721859297646e-06,
+      "loss": 0.9836,
+      "step": 2519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.643649171540883e-06,
+      "loss": 1.029,
+      "step": 2520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636431717315545e-06,
+      "loss": 1.0064,
+      "step": 2521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.629219829567149e-06,
+      "loss": 1.0525,
+      "step": 2522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6220135146042347e-06,
+      "loss": 1.0106,
+      "step": 2523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.614812778730463e-06,
+      "loss": 1.0162,
+      "step": 2524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6076176282446184e-06,
+      "loss": 1.0042,
+      "step": 2525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.600428069440598e-06,
+      "loss": 0.9902,
+      "step": 2526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593244108607403e-06,
+      "loss": 0.9881,
+      "step": 2527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5860657520291465e-06,
+      "loss": 1.0588,
+      "step": 2528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5788930059850244e-06,
+      "loss": 0.9575,
+      "step": 2529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.57172587674935e-06,
+      "loss": 0.9537,
+      "step": 2530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5645643705914978e-06,
+      "loss": 0.9946,
+      "step": 2531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5574084937759358e-06,
+      "loss": 0.9742,
+      "step": 2532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550258252562211e-06,
+      "loss": 1.0073,
+      "step": 2533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5431136532049247e-06,
+      "loss": 0.9926,
+      "step": 2534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535974701953768e-06,
+      "loss": 1.0076,
+      "step": 2535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5288414050534683e-06,
+      "loss": 1.0063,
+      "step": 2536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5217137687438208e-06,
+      "loss": 0.9916,
+      "step": 2537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5145917992596603e-06,
+      "loss": 1.0082,
+      "step": 2538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.507475502830869e-06,
+      "loss": 0.9752,
+      "step": 2539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5003648856823693e-06,
+      "loss": 1.0086,
+      "step": 2540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4932599540341107e-06,
+      "loss": 0.9744,
+      "step": 2541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4861607141010745e-06,
+      "loss": 0.9947,
+      "step": 2542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479067172093252e-06,
+      "loss": 0.9929,
+      "step": 2543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4719793342156717e-06,
+      "loss": 0.9817,
+      "step": 2544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4648972066683474e-06,
+      "loss": 1.0138,
+      "step": 2545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.457820795646316e-06,
+      "loss": 0.9696,
+      "step": 2546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4507501073396043e-06,
+      "loss": 1.0166,
+      "step": 2547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4436851479332357e-06,
+      "loss": 1.0365,
+      "step": 2548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4366259236072264e-06,
+      "loss": 0.9928,
+      "step": 2549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4295724405365693e-06,
+      "loss": 1.0452,
+      "step": 2550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4225247048912448e-06,
+      "loss": 1.0159,
+      "step": 2551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415482722836191e-06,
+      "loss": 1.0139,
+      "step": 2552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.408446500531325e-06,
+      "loss": 1.0675,
+      "step": 2553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4014160441315225e-06,
+      "loss": 1.0433,
+      "step": 2554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3943913597866172e-06,
+      "loss": 1.0044,
+      "step": 2555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3873724536413876e-06,
+      "loss": 1.0207,
+      "step": 2556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3803593318355622e-06,
+      "loss": 1.0,
+      "step": 2557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373352000503811e-06,
+      "loss": 0.9834,
+      "step": 2558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366350465775737e-06,
+      "loss": 1.0051,
+      "step": 2559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359354733775877e-06,
+      "loss": 1.0186,
+      "step": 2560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3523648106236796e-06,
+      "loss": 0.9946,
+      "step": 2561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.345380702433524e-06,
+      "loss": 0.9677,
+      "step": 2562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3384024153146986e-06,
+      "loss": 0.9889,
+      "step": 2563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3314299553714056e-06,
+      "loss": 1.0118,
+      "step": 2564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3244633287027363e-06,
+      "loss": 1.0365,
+      "step": 2565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317502541402692e-06,
+      "loss": 0.9929,
+      "step": 2566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3105475995601622e-06,
+      "loss": 1.0114,
+      "step": 2567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3035985092589206e-06,
+      "loss": 1.025,
+      "step": 2568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2966552765776304e-06,
+      "loss": 0.9818,
+      "step": 2569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2897179075898177e-06,
+      "loss": 1.0135,
+      "step": 2570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2827864083638925e-06,
+      "loss": 1.0041,
+      "step": 2571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.275860784963116e-06,
+      "loss": 1.0095,
+      "step": 2572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268941043445629e-06,
+      "loss": 1.0099,
+      "step": 2573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2620271898644095e-06,
+      "loss": 0.9681,
+      "step": 2574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255119230267292e-06,
+      "loss": 1.0106,
+      "step": 2575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.248217170696957e-06,
+      "loss": 0.9854,
+      "step": 2576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2413210171909213e-06,
+      "loss": 0.9707,
+      "step": 2577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2344307757815398e-06,
+      "loss": 0.9339,
+      "step": 2578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2275464524959856e-06,
+      "loss": 1.0164,
+      "step": 2579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.220668053356267e-06,
+      "loss": 0.9964,
+      "step": 2580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2137955843791977e-06,
+      "loss": 0.9907,
+      "step": 2581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.206929051576423e-06,
+      "loss": 0.9437,
+      "step": 2582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200068460954373e-06,
+      "loss": 1.0232,
+      "step": 2583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.193213818514295e-06,
+      "loss": 1.0097,
+      "step": 2584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186365130252231e-06,
+      "loss": 0.957,
+      "step": 2585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.179522402159004e-06,
+      "loss": 1.015,
+      "step": 2586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1726856402202444e-06,
+      "loss": 1.0404,
+      "step": 2587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.165854850416342e-06,
+      "loss": 0.3984,
+      "step": 2588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1590300387224772e-06,
+      "loss": 0.9683,
+      "step": 2589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1522112111085902e-06,
+      "loss": 0.9708,
+      "step": 2590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1453983735393943e-06,
+      "loss": 0.9607,
+      "step": 2591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.138591531974362e-06,
+      "loss": 0.9896,
+      "step": 2592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.13179069236772e-06,
+      "loss": 1.0195,
+      "step": 2593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.124995860668446e-06,
+      "loss": 1.0251,
+      "step": 2594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1182070428202536e-06,
+      "loss": 0.962,
+      "step": 2595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1114242447616137e-06,
+      "loss": 1.043,
+      "step": 2596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104647472425714e-06,
+      "loss": 1.0184,
+      "step": 2597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0978767317404813e-06,
+      "loss": 1.0317,
+      "step": 2598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091112028628559e-06,
+      "loss": 0.9433,
+      "step": 2599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0843533690073156e-06,
+      "loss": 0.964,
+      "step": 2600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0776007587888324e-06,
+      "loss": 0.397,
+      "step": 2601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070854203879895e-06,
+      "loss": 0.9425,
+      "step": 2602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0641137101819994e-06,
+      "loss": 1.0015,
+      "step": 2603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0573792835913273e-06,
+      "loss": 1.0147,
+      "step": 2604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0506509299987662e-06,
+      "loss": 0.9919,
+      "step": 2605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.043928655289884e-06,
+      "loss": 1.0002,
+      "step": 2606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0372124653449352e-06,
+      "loss": 1.0027,
+      "step": 2607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0305023660388457e-06,
+      "loss": 0.3601,
+      "step": 2608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.023798363241219e-06,
+      "loss": 0.9825,
+      "step": 2609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017100462816326e-06,
+      "loss": 1.0628,
+      "step": 2610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010408670623096e-06,
+      "loss": 1.0659,
+      "step": 2611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0037229925151202e-06,
+      "loss": 0.9648,
+      "step": 2612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997043434340633e-06,
+      "loss": 0.9953,
+      "step": 2613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.990370001942523e-06,
+      "loss": 1.0094,
+      "step": 2614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9837027011583196e-06,
+      "loss": 1.0112,
+      "step": 2615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.977041537820189e-06,
+      "loss": 0.9889,
+      "step": 2616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9703865177549207e-06,
+      "loss": 1.0145,
+      "step": 2617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9637376467839405e-06,
+      "loss": 0.9526,
+      "step": 2618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.957094930723291e-06,
+      "loss": 0.9675,
+      "step": 2619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9504583753836304e-06,
+      "loss": 0.9749,
+      "step": 2620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9438279865702333e-06,
+      "loss": 1.0094,
+      "step": 2621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9372037700829693e-06,
+      "loss": 0.9843,
+      "step": 2622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.93058573171632e-06,
+      "loss": 1.0101,
+      "step": 2623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9239738772593495e-06,
+      "loss": 1.0445,
+      "step": 2624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9173682124957325e-06,
+      "loss": 0.9945,
+      "step": 2625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.91076874320371e-06,
+      "loss": 1.0011,
+      "step": 2626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9041754751561134e-06,
+      "loss": 1.0177,
+      "step": 2627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897588414120349e-06,
+      "loss": 0.9487,
+      "step": 2628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8910075658583893e-06,
+      "loss": 1.0296,
+      "step": 2629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884432936126781e-06,
+      "loss": 0.9786,
+      "step": 2630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8778645306766174e-06,
+      "loss": 1.0004,
+      "step": 2631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.871302355253561e-06,
+      "loss": 0.9474,
+      "step": 2632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.86474641559781e-06,
+      "loss": 0.9755,
+      "step": 2633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8581967174441293e-06,
+      "loss": 0.9756,
+      "step": 2634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851653266521802e-06,
+      "loss": 1.0331,
+      "step": 2635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8451160685546587e-06,
+      "loss": 1.009,
+      "step": 2636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838585129261059e-06,
+      "loss": 1.0268,
+      "step": 2637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.83206045435388e-06,
+      "loss": 1.0216,
+      "step": 2638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825542049540535e-06,
+      "loss": 0.9879,
+      "step": 2639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8190299205229354e-06,
+      "loss": 0.95,
+      "step": 2640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8125240729975145e-06,
+      "loss": 1.0183,
+      "step": 2641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806024512655203e-06,
+      "loss": 1.0223,
+      "step": 2642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.799531245181437e-06,
+      "loss": 1.0337,
+      "step": 2643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.793044276256148e-06,
+      "loss": 0.9468,
+      "step": 2644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7865636115537566e-06,
+      "loss": 0.9814,
+      "step": 2645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7800892567431705e-06,
+      "loss": 0.9934,
+      "step": 2646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7736212174877675e-06,
+      "loss": 0.9643,
+      "step": 2647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.767159499445423e-06,
+      "loss": 0.9878,
+      "step": 2648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7607041082684592e-06,
+      "loss": 0.9969,
+      "step": 2649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.754255049603681e-06,
+      "loss": 0.9933,
+      "step": 2650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.747812329092341e-06,
+      "loss": 0.9937,
+      "step": 2651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7413759523701578e-06,
+      "loss": 1.0261,
+      "step": 2652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.734945925067295e-06,
+      "loss": 1.0212,
+      "step": 2653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7285222528083655e-06,
+      "loss": 1.0313,
+      "step": 2654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7221049412124246e-06,
+      "loss": 1.023,
+      "step": 2655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.715693995892954e-06,
+      "loss": 1.0008,
+      "step": 2656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.709289422457877e-06,
+      "loss": 0.3899,
+      "step": 2657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7028912265095396e-06,
+      "loss": 1.043,
+      "step": 2658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696499413644712e-06,
+      "loss": 0.994,
+      "step": 2659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69011398945457e-06,
+      "loss": 0.9953,
+      "step": 2660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6837349595247154e-06,
+      "loss": 1.0447,
+      "step": 2661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677362329435146e-06,
+      "loss": 0.9829,
+      "step": 2662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6709961047602686e-06,
+      "loss": 1.0002,
+      "step": 2663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.664636291068885e-06,
+      "loss": 1.049,
+      "step": 2664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6582828939241834e-06,
+      "loss": 1.002,
+      "step": 2665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6519359188837447e-06,
+      "loss": 0.9875,
+      "step": 2666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.645595371499533e-06,
+      "loss": 0.9912,
+      "step": 2667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.63926125731789e-06,
+      "loss": 0.9731,
+      "step": 2668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.632933581879521e-06,
+      "loss": 0.9991,
+      "step": 2669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6266123507195105e-06,
+      "loss": 1.0214,
+      "step": 2670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6202975693673007e-06,
+      "loss": 0.3782,
+      "step": 2671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613989243346693e-06,
+      "loss": 1.0139,
+      "step": 2672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607687378175844e-06,
+      "loss": 1.0159,
+      "step": 2673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6013919793672515e-06,
+      "loss": 1.0043,
+      "step": 2674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5951030524277665e-06,
+      "loss": 0.9757,
+      "step": 2675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588820602858565e-06,
+      "loss": 0.9527,
+      "step": 2676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5825446361551785e-06,
+      "loss": 1.0286,
+      "step": 2677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5762751578074475e-06,
+      "loss": 0.9756,
+      "step": 2678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5700121732995464e-06,
+      "loss": 0.9864,
+      "step": 2679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563755688109969e-06,
+      "loss": 0.9668,
+      "step": 2680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5575057077115163e-06,
+      "loss": 1.025,
+      "step": 2681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5512622375713146e-06,
+      "loss": 0.3532,
+      "step": 2682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545025283150777e-06,
+      "loss": 1.0057,
+      "step": 2683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.538794849905635e-06,
+      "loss": 0.98,
+      "step": 2684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5325709432858946e-06,
+      "loss": 0.9823,
+      "step": 2685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5263535687358774e-06,
+      "loss": 0.9912,
+      "step": 2686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.520142731694172e-06,
+      "loss": 0.9539,
+      "step": 2687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5139384375936547e-06,
+      "loss": 0.9737,
+      "step": 2688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5077406918614842e-06,
+      "loss": 0.9701,
+      "step": 2689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.501549499919076e-06,
+      "loss": 0.9761,
+      "step": 2690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.495364867182133e-06,
+      "loss": 0.9782,
+      "step": 2691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489186799060602e-06,
+      "loss": 1.0165,
+      "step": 2692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.483015300958702e-06,
+      "loss": 1.0417,
+      "step": 2693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.476850378274891e-06,
+      "loss": 1.0212,
+      "step": 2694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4706920364018848e-06,
+      "loss": 1.0074,
+      "step": 2695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4645402807266428e-06,
+      "loss": 0.9915,
+      "step": 2696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.458395116630358e-06,
+      "loss": 0.9978,
+      "step": 2697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4522565494884645e-06,
+      "loss": 1.0502,
+      "step": 2698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4461245846706116e-06,
+      "loss": 0.995,
+      "step": 2699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439999227540696e-06,
+      "loss": 0.3324,
+      "step": 2700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.433880483456812e-06,
+      "loss": 0.3734,
+      "step": 2701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.427768357771285e-06,
+      "loss": 0.3608,
+      "step": 2702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.421662855830638e-06,
+      "loss": 0.9965,
+      "step": 2703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4155639829756106e-06,
+      "loss": 1.0087,
+      "step": 2704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409471744541139e-06,
+      "loss": 1.039,
+      "step": 2705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.403386145856357e-06,
+      "loss": 1.0181,
+      "step": 2706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3973071922445924e-06,
+      "loss": 1.0119,
+      "step": 2707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3912348890233526e-06,
+      "loss": 1.0002,
+      "step": 2708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.385169241504336e-06,
+      "loss": 0.9966,
+      "step": 2709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379110254993414e-06,
+      "loss": 0.9674,
+      "step": 2710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3730579347906403e-06,
+      "loss": 0.9673,
+      "step": 2711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.36701228619022e-06,
+      "loss": 0.9888,
+      "step": 2712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360973314480538e-06,
+      "loss": 0.9899,
+      "step": 2713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354941024944132e-06,
+      "loss": 1.0173,
+      "step": 2714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.348915422857696e-06,
+      "loss": 0.3444,
+      "step": 2715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.342896513492078e-06,
+      "loss": 1.0011,
+      "step": 2716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33688430211226e-06,
+      "loss": 0.9886,
+      "step": 2717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3308787939773757e-06,
+      "loss": 1.0264,
+      "step": 2718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.324879994340693e-06,
+      "loss": 1.0176,
+      "step": 2719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.318887908449612e-06,
+      "loss": 0.9856,
+      "step": 2720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312902541545654e-06,
+      "loss": 1.0243,
+      "step": 2721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3069238988644704e-06,
+      "loss": 1.0607,
+      "step": 2722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3009519856358277e-06,
+      "loss": 1.0529,
+      "step": 2723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2949868070836044e-06,
+      "loss": 1.0228,
+      "step": 2724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2890283684257954e-06,
+      "loss": 0.9677,
+      "step": 2725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283076674874485e-06,
+      "loss": 1.0051,
+      "step": 2726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2771317316358733e-06,
+      "loss": 1.0368,
+      "step": 2727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2711935439102406e-06,
+      "loss": 0.9961,
+      "step": 2728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2652621168919765e-06,
+      "loss": 0.9233,
+      "step": 2729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.259337455769538e-06,
+      "loss": 1.0146,
+      "step": 2730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2534195657254744e-06,
+      "loss": 0.9927,
+      "step": 2731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247508451936411e-06,
+      "loss": 0.9867,
+      "step": 2732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2416041195730378e-06,
+      "loss": 1.0163,
+      "step": 2733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2357065738001294e-06,
+      "loss": 0.9363,
+      "step": 2734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229815819776504e-06,
+      "loss": 0.9774,
+      "step": 2735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2239318626550565e-06,
+      "loss": 0.9823,
+      "step": 2736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2180547075827164e-06,
+      "loss": 0.9888,
+      "step": 2737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212184359700489e-06,
+      "loss": 0.3458,
+      "step": 2738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2063208241434007e-06,
+      "loss": 0.9837,
+      "step": 2739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2004641060405318e-06,
+      "loss": 0.9634,
+      "step": 2740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1946142105150013e-06,
+      "loss": 0.9809,
+      "step": 2741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188771142683944e-06,
+      "loss": 1.0115,
+      "step": 2742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.182934907658547e-06,
+      "loss": 1.001,
+      "step": 2743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1771055105439973e-06,
+      "loss": 1.0045,
+      "step": 2744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171282956439517e-06,
+      "loss": 1.0096,
+      "step": 2745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165467250438328e-06,
+      "loss": 0.9794,
+      "step": 2746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159658397627673e-06,
+      "loss": 1.0037,
+      "step": 2747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1538564030887976e-06,
+      "loss": 0.9486,
+      "step": 2748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.148061271896946e-06,
+      "loss": 0.993,
+      "step": 2749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1422730091213617e-06,
+      "loss": 0.9725,
+      "step": 2750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136491619825272e-06,
+      "loss": 0.9745,
+      "step": 2751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.130717109065905e-06,
+      "loss": 1.0253,
+      "step": 2752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1249494818944604e-06,
+      "loss": 0.9917,
+      "step": 2753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1191887433561243e-06,
+      "loss": 0.356,
+      "step": 2754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1134348984900477e-06,
+      "loss": 1.0262,
+      "step": 2755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1076879523293604e-06,
+      "loss": 1.029,
+      "step": 2756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1019479099011543e-06,
+      "loss": 0.9874,
+      "step": 2757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0962147762264815e-06,
+      "loss": 1.0231,
+      "step": 2758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090488556320357e-06,
+      "loss": 0.98,
+      "step": 2759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0847692551917355e-06,
+      "loss": 0.9851,
+      "step": 2760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0790568778435295e-06,
+      "loss": 0.3636,
+      "step": 2761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.073351429272592e-06,
+      "loss": 1.0124,
+      "step": 2762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067652914469719e-06,
+      "loss": 1.0309,
+      "step": 2763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061961338419631e-06,
+      "loss": 1.0212,
+      "step": 2764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0562767061009893e-06,
+      "loss": 0.9764,
+      "step": 2765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.050599022486377e-06,
+      "loss": 0.9305,
+      "step": 2766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0449282925422987e-06,
+      "loss": 0.9917,
+      "step": 2767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0392645212291805e-06,
+      "loss": 0.9829,
+      "step": 2768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033607713501353e-06,
+      "loss": 1.0271,
+      "step": 2769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0279578743070615e-06,
+      "loss": 0.9928,
+      "step": 2770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0223150085884557e-06,
+      "loss": 1.0186,
+      "step": 2771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016679121281586e-06,
+      "loss": 1.007,
+      "step": 2772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0110502173163916e-06,
+      "loss": 0.9662,
+      "step": 2773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0054283016167107e-06,
+      "loss": 1.0071,
+      "step": 2774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9998133791002674e-06,
+      "loss": 0.9734,
+      "step": 2775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994205454678665e-06,
+      "loss": 0.999,
+      "step": 2776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988604533257391e-06,
+      "loss": 1.0336,
+      "step": 2777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983010619735799e-06,
+      "loss": 1.0052,
+      "step": 2778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977423719007121e-06,
+      "loss": 0.9564,
+      "step": 2779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9718438359584434e-06,
+      "loss": 0.9955,
+      "step": 2780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966270975470732e-06,
+      "loss": 0.3661,
+      "step": 2781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960705142418791e-06,
+      "loss": 0.9742,
+      "step": 2782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9551463416712867e-06,
+      "loss": 0.9854,
+      "step": 2783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9495945780907367e-06,
+      "loss": 0.9984,
+      "step": 2784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944049856533489e-06,
+      "loss": 0.9579,
+      "step": 2785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9385121818497534e-06,
+      "loss": 0.9897,
+      "step": 2786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9329815588835555e-06,
+      "loss": 1.0242,
+      "step": 2787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9274579924727653e-06,
+      "loss": 1.0449,
+      "step": 2788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9219414874490672e-06,
+      "loss": 0.9615,
+      "step": 2789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9164320486379884e-06,
+      "loss": 0.9999,
+      "step": 2790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910929680858854e-06,
+      "loss": 0.9731,
+      "step": 2791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9054343889248162e-06,
+      "loss": 0.9948,
+      "step": 2792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8999461776428384e-06,
+      "loss": 0.991,
+      "step": 2793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8944650518136776e-06,
+      "loss": 1.0037,
+      "step": 2794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8889910162319113e-06,
+      "loss": 0.9839,
+      "step": 2795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8835240756858996e-06,
+      "loss": 0.9715,
+      "step": 2796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.878064234957806e-06,
+      "loss": 0.9531,
+      "step": 2797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8726114988235755e-06,
+      "loss": 0.9958,
+      "step": 2798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.867165872052943e-06,
+      "loss": 0.933,
+      "step": 2799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8617273594094243e-06,
+      "loss": 1.0178,
+      "step": 2800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8562959656503131e-06,
+      "loss": 0.9969,
+      "step": 2801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.850871695526677e-06,
+      "loss": 0.325,
+      "step": 2802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8454545537833447e-06,
+      "loss": 1.0176,
+      "step": 2803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8400445451589166e-06,
+      "loss": 0.9939,
+      "step": 2804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8346416743857508e-06,
+      "loss": 1.0119,
+      "step": 2805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.829245946189967e-06,
+      "loss": 0.9938,
+      "step": 2806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.823857365291426e-06,
+      "loss": 0.9942,
+      "step": 2807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8184759364037452e-06,
+      "loss": 0.9715,
+      "step": 2808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8131016642342835e-06,
+      "loss": 0.9404,
+      "step": 2809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8077345534841394e-06,
+      "loss": 1.0293,
+      "step": 2810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.802374608848151e-06,
+      "loss": 0.9598,
+      "step": 2811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.797021835014877e-06,
+      "loss": 0.9767,
+      "step": 2812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7916762366666153e-06,
+      "loss": 1.0684,
+      "step": 2813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7863378184793801e-06,
+      "loss": 0.3304,
+      "step": 2814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.781006585122912e-06,
+      "loss": 0.9598,
+      "step": 2815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7756825412606537e-06,
+      "loss": 0.9434,
+      "step": 2816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7703656915497724e-06,
+      "loss": 1.0396,
+      "step": 2817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7650560406411354e-06,
+      "loss": 0.3518,
+      "step": 2818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.759753593179314e-06,
+      "loss": 0.9941,
+      "step": 2819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.754458353802584e-06,
+      "loss": 0.9567,
+      "step": 2820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7491703271429029e-06,
+      "loss": 0.9617,
+      "step": 2821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7438895178259297e-06,
+      "loss": 0.9972,
+      "step": 2822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7386159304710083e-06,
+      "loss": 0.9855,
+      "step": 2823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.733349569691166e-06,
+      "loss": 1.0109,
+      "step": 2824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.728090440093102e-06,
+      "loss": 0.9801,
+      "step": 2825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7228385462771967e-06,
+      "loss": 0.9638,
+      "step": 2826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7175938928375012e-06,
+      "loss": 1.0085,
+      "step": 2827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.71235648436173e-06,
+      "loss": 0.9743,
+      "step": 2828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707126325431263e-06,
+      "loss": 0.995,
+      "step": 2829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019034206211338e-06,
+      "loss": 0.9866,
+      "step": 2830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6966877745000375e-06,
+      "loss": 0.9946,
+      "step": 2831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6914793916303095e-06,
+      "loss": 0.9868,
+      "step": 2832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6862782765679475e-06,
+      "loss": 1.0174,
+      "step": 2833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6810844338625742e-06,
+      "loss": 1.0133,
+      "step": 2834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6758978680574623e-06,
+      "loss": 1.045,
+      "step": 2835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670718583689519e-06,
+      "loss": 0.9748,
+      "step": 2836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6655465852892705e-06,
+      "loss": 0.9888,
+      "step": 2837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660381877380889e-06,
+      "loss": 1.0265,
+      "step": 2838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6552244644821503e-06,
+      "loss": 1.0072,
+      "step": 2839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6500743511044636e-06,
+      "loss": 1.0153,
+      "step": 2840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6449315417528377e-06,
+      "loss": 0.9851,
+      "step": 2841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6397960409259105e-06,
+      "loss": 0.9786,
+      "step": 2842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6346678531159122e-06,
+      "loss": 1.027,
+      "step": 2843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6295469828086818e-06,
+      "loss": 1.0022,
+      "step": 2844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6244334344836598e-06,
+      "loss": 0.97,
+      "step": 2845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6193272126138727e-06,
+      "loss": 1.034,
+      "step": 2846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6142283216659526e-06,
+      "loss": 1.0212,
+      "step": 2847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6091367661001033e-06,
+      "loss": 1.02,
+      "step": 2848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6040525503701255e-06,
+      "loss": 0.9993,
+      "step": 2849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598975678923389e-06,
+      "loss": 0.9583,
+      "step": 2850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.593906156200844e-06,
+      "loss": 0.9677,
+      "step": 2851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5888439866370141e-06,
+      "loss": 1.0233,
+      "step": 2852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5837891746599887e-06,
+      "loss": 1.0069,
+      "step": 2853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5787417246914228e-06,
+      "loss": 0.987,
+      "step": 2854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5737016411465257e-06,
+      "loss": 0.9608,
+      "step": 2855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5686689284340695e-06,
+      "loss": 1.0151,
+      "step": 2856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5636435909563763e-06,
+      "loss": 0.9555,
+      "step": 2857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5586256331093207e-06,
+      "loss": 0.9833,
+      "step": 2858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.553615059282314e-06,
+      "loss": 0.9735,
+      "step": 2859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5486118738583134e-06,
+      "loss": 0.9667,
+      "step": 2860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5436160812138135e-06,
+      "loss": 0.9773,
+      "step": 2861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538627685718841e-06,
+      "loss": 1.0134,
+      "step": 2862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5336466917369552e-06,
+      "loss": 1.0062,
+      "step": 2863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5286731036252323e-06,
+      "loss": 0.9978,
+      "step": 2864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5237069257342784e-06,
+      "loss": 0.9946,
+      "step": 2865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5187481624082167e-06,
+      "loss": 0.9772,
+      "step": 2866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513796817984684e-06,
+      "loss": 1.0228,
+      "step": 2867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5088528967948213e-06,
+      "loss": 0.369,
+      "step": 2868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5039164031632847e-06,
+      "loss": 0.9379,
+      "step": 2869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.498987341408229e-06,
+      "loss": 0.9806,
+      "step": 2870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4940657158413109e-06,
+      "loss": 0.9829,
+      "step": 2871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4891515307676795e-06,
+      "loss": 0.9796,
+      "step": 2872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4842447904859737e-06,
+      "loss": 0.9692,
+      "step": 2873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4793454992883249e-06,
+      "loss": 0.9855,
+      "step": 2874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4744536614603454e-06,
+      "loss": 1.0472,
+      "step": 2875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4695692812811302e-06,
+      "loss": 1.0125,
+      "step": 2876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4646923630232456e-06,
+      "loss": 0.9813,
+      "step": 2877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4598229109527372e-06,
+      "loss": 0.9971,
+      "step": 2878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4549609293291145e-06,
+      "loss": 0.9558,
+      "step": 2879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4501064224053562e-06,
+      "loss": 0.9551,
+      "step": 2880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4452593944279014e-06,
+      "loss": 0.9825,
+      "step": 2881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4404198496366428e-06,
+      "loss": 0.9896,
+      "step": 2882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.435587792264933e-06,
+      "loss": 0.9725,
+      "step": 2883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4307632265395733e-06,
+      "loss": 0.9617,
+      "step": 2884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.425946156680813e-06,
+      "loss": 0.9792,
+      "step": 2885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4211365869023374e-06,
+      "loss": 0.9586,
+      "step": 2886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.41633452141128e-06,
+      "loss": 0.9588,
+      "step": 2887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4115399644082096e-06,
+      "loss": 0.9883,
+      "step": 2888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4067529200871156e-06,
+      "loss": 0.9421,
+      "step": 2889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4019733926354328e-06,
+      "loss": 0.972,
+      "step": 2890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.397201386234006e-06,
+      "loss": 0.9667,
+      "step": 2891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3924369050571097e-06,
+      "loss": 0.9872,
+      "step": 2892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3876799532724328e-06,
+      "loss": 0.9696,
+      "step": 2893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3829305350410805e-06,
+      "loss": 0.9795,
+      "step": 2894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3781886545175606e-06,
+      "loss": 1.0196,
+      "step": 2895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373454315849796e-06,
+      "loss": 0.9467,
+      "step": 2896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3687275231791097e-06,
+      "loss": 1.0069,
+      "step": 2897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3640082806402167e-06,
+      "loss": 0.9923,
+      "step": 2898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3592965923612445e-06,
+      "loss": 0.9866,
+      "step": 2899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.354592462463694e-06,
+      "loss": 1.0014,
+      "step": 2900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3498958950624653e-06,
+      "loss": 1.0163,
+      "step": 2901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3452068942658437e-06,
+      "loss": 1.039,
+      "step": 2902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3405254641754861e-06,
+      "loss": 0.9877,
+      "step": 2903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3358516088864371e-06,
+      "loss": 0.9691,
+      "step": 2904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3311853324871115e-06,
+      "loss": 1.0006,
+      "step": 2905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3265266390592956e-06,
+      "loss": 0.3598,
+      "step": 2906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3218755326781396e-06,
+      "loss": 0.9583,
+      "step": 2907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.31723201741216e-06,
+      "loss": 0.3393,
+      "step": 2908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.312596097323231e-06,
+      "loss": 1.0082,
+      "step": 2909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3079677764665866e-06,
+      "loss": 1.016,
+      "step": 2910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3033470588908105e-06,
+      "loss": 0.9631,
+      "step": 2911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2987339486378337e-06,
+      "loss": 0.986,
+      "step": 2912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2941284497429351e-06,
+      "loss": 1.0409,
+      "step": 2913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2895305662347368e-06,
+      "loss": 0.9982,
+      "step": 2914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2849403021351992e-06,
+      "loss": 0.9614,
+      "step": 2915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2803576614596135e-06,
+      "loss": 1.0176,
+      "step": 2916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.275782648216607e-06,
+      "loss": 1.0274,
+      "step": 2917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2712152664081334e-06,
+      "loss": 0.9306,
+      "step": 2918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2666555200294705e-06,
+      "loss": 0.9929,
+      "step": 2919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2621034130692212e-06,
+      "loss": 1.0213,
+      "step": 2920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2575589495092954e-06,
+      "loss": 0.9813,
+      "step": 2921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2530221333249282e-06,
+      "loss": 0.9932,
+      "step": 2922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2484929684846592e-06,
+      "loss": 0.3236,
+      "step": 2923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2439714589503393e-06,
+      "loss": 0.9841,
+      "step": 2924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.239457608677117e-06,
+      "loss": 1.0196,
+      "step": 2925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.234951421613444e-06,
+      "loss": 0.9914,
+      "step": 2926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2304529017010714e-06,
+      "loss": 1.0055,
+      "step": 2927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2259620528750394e-06,
+      "loss": 0.9658,
+      "step": 2928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2214788790636823e-06,
+      "loss": 0.991,
+      "step": 2929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2170033841886143e-06,
+      "loss": 1.0248,
+      "step": 2930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.212535572164737e-06,
+      "loss": 1.0241,
+      "step": 2931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2080754469002321e-06,
+      "loss": 1.0212,
+      "step": 2932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2036230122965575e-06,
+      "loss": 0.9781,
+      "step": 2933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.199178272248438e-06,
+      "loss": 0.9918,
+      "step": 2934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1947412306438745e-06,
+      "loss": 0.9942,
+      "step": 2935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1903118913641309e-06,
+      "loss": 0.9814,
+      "step": 2936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185890258283735e-06,
+      "loss": 1.0124,
+      "step": 2937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1814763352704727e-06,
+      "loss": 0.9665,
+      "step": 2938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1770701261853822e-06,
+      "loss": 0.9577,
+      "step": 2939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1726716348827617e-06,
+      "loss": 1.0118,
+      "step": 2940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1682808652101474e-06,
+      "loss": 0.9876,
+      "step": 2941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1638978210083353e-06,
+      "loss": 0.9715,
+      "step": 2942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1595225061113503e-06,
+      "loss": 0.9461,
+      "step": 2943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1551549243464633e-06,
+      "loss": 1.0019,
+      "step": 2944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1507950795341793e-06,
+      "loss": 1.0034,
+      "step": 2945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1464429754882356e-06,
+      "loss": 0.9718,
+      "step": 2946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1420986160156e-06,
+      "loss": 0.9938,
+      "step": 2947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1377620049164585e-06,
+      "loss": 0.9991,
+      "step": 2948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1334331459842295e-06,
+      "loss": 1.0018,
+      "step": 2949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1291120430055403e-06,
+      "loss": 0.9712,
+      "step": 2950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1247986997602456e-06,
+      "loss": 1.0191,
+      "step": 2951
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1204931200213998e-06,
+      "loss": 0.9895,
+      "step": 2952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1161953075552733e-06,
+      "loss": 0.9792,
+      "step": 2953
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1119052661213435e-06,
+      "loss": 0.3714,
+      "step": 2954
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.107622999472282e-06,
+      "loss": 0.9993,
+      "step": 2955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.103348511353972e-06,
+      "loss": 0.9905,
+      "step": 2956
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0990818055054808e-06,
+      "loss": 0.999,
+      "step": 2957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0948228856590748e-06,
+      "loss": 1.0246,
+      "step": 2958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0905717555402051e-06,
+      "loss": 0.9839,
+      "step": 2959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0863284188675128e-06,
+      "loss": 0.9824,
+      "step": 2960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0820928793528207e-06,
+      "loss": 1.0113,
+      "step": 2961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.077865140701131e-06,
+      "loss": 0.988,
+      "step": 2962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.073645206610624e-06,
+      "loss": 0.9361,
+      "step": 2963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0694330807726438e-06,
+      "loss": 1.0258,
+      "step": 2964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.065228766871721e-06,
+      "loss": 0.9785,
+      "step": 2965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0610322685855357e-06,
+      "loss": 0.9955,
+      "step": 2966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0568435895849449e-06,
+      "loss": 0.9877,
+      "step": 2967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0526627335339534e-06,
+      "loss": 0.927,
+      "step": 2968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0484897040897335e-06,
+      "loss": 0.9794,
+      "step": 2969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0443245049026051e-06,
+      "loss": 0.9867,
+      "step": 2970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040167139616043e-06,
+      "loss": 1.0538,
+      "step": 2971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0360176118666654e-06,
+      "loss": 1.0187,
+      "step": 2972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.031875925284236e-06,
+      "loss": 1.0047,
+      "step": 2973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0277420834916586e-06,
+      "loss": 1.0161,
+      "step": 2974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0236160901049774e-06,
+      "loss": 0.963,
+      "step": 2975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0194979487333723e-06,
+      "loss": 0.9823,
+      "step": 2976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.015387662979148e-06,
+      "loss": 1.0114,
+      "step": 2977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0112852364377424e-06,
+      "loss": 1.0302,
+      "step": 2978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0071906726977198e-06,
+      "loss": 0.9916,
+      "step": 2979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0031039753407646e-06,
+      "loss": 0.935,
+      "step": 2980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.990251479416812e-07,
+      "loss": 1.0023,
+      "step": 2981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.94954194068386e-07,
+      "loss": 0.9788,
+      "step": 2982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.908911172819135e-07,
+      "loss": 1.0065,
+      "step": 2983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.868359211364032e-07,
+      "loss": 0.9607,
+      "step": 2984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827886091791083e-07,
+      "loss": 0.9829,
+      "step": 2985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.787491849503739e-07,
+      "loss": 1.0239,
+      "step": 2986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.747176519836542e-07,
+      "loss": 1.0315,
+      "step": 2987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.706940138054998e-07,
+      "loss": 0.9553,
+      "step": 2988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.666782739355528e-07,
+      "loss": 1.0007,
+      "step": 2989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62670435886548e-07,
+      "loss": 1.0204,
+      "step": 2990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.586705031643052e-07,
+      "loss": 0.9932,
+      "step": 2991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.546784792677356e-07,
+      "loss": 0.9581,
+      "step": 2992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.506943676888203e-07,
+      "loss": 1.008,
+      "step": 2993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.467181719126361e-07,
+      "loss": 0.988,
+      "step": 2994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.427498954173197e-07,
+      "loss": 0.9736,
+      "step": 2995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.387895416740889e-07,
+      "loss": 1.0109,
+      "step": 2996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.348371141472301e-07,
+      "loss": 1.0141,
+      "step": 2997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.308926162940946e-07,
+      "loss": 1.0118,
+      "step": 2998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.269560515650999e-07,
+      "loss": 1.0079,
+      "step": 2999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.230274234037206e-07,
+      "loss": 0.9786,
+      "step": 3000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.19106735246491e-07,
+      "loss": 0.9828,
+      "step": 3001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151939905229978e-07,
+      "loss": 0.9891,
+      "step": 3002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.112891926558854e-07,
+      "loss": 1.0009,
+      "step": 3003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.073923450608368e-07,
+      "loss": 1.0133,
+      "step": 3004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.035034511465901e-07,
+      "loss": 0.9688,
+      "step": 3005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.996225143149229e-07,
+      "loss": 1.0018,
+      "step": 3006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.957495379606451e-07,
+      "loss": 0.988,
+      "step": 3007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.91884525471618e-07,
+      "loss": 0.9871,
+      "step": 3008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.880274802287226e-07,
+      "loss": 1.0497,
+      "step": 3009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.841784056058789e-07,
+      "loss": 0.9857,
+      "step": 3010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.803373049700304e-07,
+      "loss": 0.9978,
+      "step": 3011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.765041816811471e-07,
+      "loss": 0.9849,
+      "step": 3012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.726790390922213e-07,
+      "loss": 0.9469,
+      "step": 3013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.688618805492632e-07,
+      "loss": 0.9619,
+      "step": 3014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.65052709391303e-07,
+      "loss": 0.954,
+      "step": 3015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.612515289503742e-07,
+      "loss": 0.9863,
+      "step": 3016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.574583425515337e-07,
+      "loss": 0.341,
+      "step": 3017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.53673153512834e-07,
+      "loss": 1.0421,
+      "step": 3018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.498959651453409e-07,
+      "loss": 0.9498,
+      "step": 3019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.461267807531126e-07,
+      "loss": 0.9783,
+      "step": 3020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.423656036332139e-07,
+      "loss": 0.9912,
+      "step": 3021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.386124370757021e-07,
+      "loss": 0.9731,
+      "step": 3022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.348672843636274e-07,
+      "loss": 0.9263,
+      "step": 3023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.311301487730317e-07,
+      "loss": 0.9774,
+      "step": 3024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.274010335729399e-07,
+      "loss": 1.0291,
+      "step": 3025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.236799420253638e-07,
+      "loss": 0.9703,
+      "step": 3026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.199668773852976e-07,
+      "loss": 0.9603,
+      "step": 3027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.162618429007141e-07,
+      "loss": 1.0074,
+      "step": 3028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.125648418125575e-07,
+      "loss": 0.9313,
+      "step": 3029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.088758773547489e-07,
+      "loss": 0.9713,
+      "step": 3030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.051949527541791e-07,
+      "loss": 0.9903,
+      "step": 3031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.015220712307059e-07,
+      "loss": 0.3357,
+      "step": 3032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.978572359971504e-07,
+      "loss": 1.0006,
+      "step": 3033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.942004502592948e-07,
+      "loss": 1.0084,
+      "step": 3034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.905517172158827e-07,
+      "loss": 0.9811,
+      "step": 3035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.86911040058611e-07,
+      "loss": 1.0159,
+      "step": 3036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.832784219721324e-07,
+      "loss": 0.9325,
+      "step": 3037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.79653866134047e-07,
+      "loss": 1.0348,
+      "step": 3038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.760373757149042e-07,
+      "loss": 0.9796,
+      "step": 3039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.724289538781982e-07,
+      "loss": 1.019,
+      "step": 3040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.688286037803649e-07,
+      "loss": 0.9723,
+      "step": 3041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.652363285707832e-07,
+      "loss": 0.3305,
+      "step": 3042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.616521313917602e-07,
+      "loss": 0.968,
+      "step": 3043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.580760153785449e-07,
+      "loss": 0.9585,
+      "step": 3044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.545079836593095e-07,
+      "loss": 0.9943,
+      "step": 3045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.509480393551671e-07,
+      "loss": 1.0039,
+      "step": 3046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.473961855801415e-07,
+      "loss": 1.0208,
+      "step": 3047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.438524254411894e-07,
+      "loss": 0.9902,
+      "step": 3048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.403167620381868e-07,
+      "loss": 0.9958,
+      "step": 3049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.367891984639186e-07,
+      "loss": 0.9765,
+      "step": 3050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.332697378040987e-07,
+      "loss": 0.9851,
+      "step": 3051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.29758383137339e-07,
+      "loss": 1.0081,
+      "step": 3052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.262551375351723e-07,
+      "loss": 0.9732,
+      "step": 3053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.227600040620253e-07,
+      "loss": 1.0101,
+      "step": 3054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.192729857752446e-07,
+      "loss": 1.0079,
+      "step": 3055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.157940857250645e-07,
+      "loss": 0.9764,
+      "step": 3056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123233069546232e-07,
+      "loss": 0.9726,
+      "step": 3057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.088606524999575e-07,
+      "loss": 0.9885,
+      "step": 3058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.054061253899913e-07,
+      "loss": 1.0429,
+      "step": 3059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.019597286465462e-07,
+      "loss": 0.9756,
+      "step": 3060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.985214652843252e-07,
+      "loss": 1.0042,
+      "step": 3061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950913383109237e-07,
+      "loss": 0.9792,
+      "step": 3062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.916693507268101e-07,
+      "loss": 1.0205,
+      "step": 3063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.882555055253437e-07,
+      "loss": 0.999,
+      "step": 3064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.848498056927532e-07,
+      "loss": 1.0313,
+      "step": 3065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.81452254208148e-07,
+      "loss": 0.9819,
+      "step": 3066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.780628540435075e-07,
+      "loss": 1.0012,
+      "step": 3067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.746816081636765e-07,
+      "loss": 1.0108,
+      "step": 3068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.713085195263757e-07,
+      "loss": 1.0038,
+      "step": 3069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.679435910821818e-07,
+      "loss": 1.0185,
+      "step": 3070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.645868257745413e-07,
+      "loss": 1.0122,
+      "step": 3071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.612382265397521e-07,
+      "loss": 0.3495,
+      "step": 3072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.578977963069732e-07,
+      "loss": 1.0477,
+      "step": 3073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.545655379982186e-07,
+      "loss": 0.3376,
+      "step": 3074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.512414545283529e-07,
+      "loss": 0.997,
+      "step": 3075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.479255488050906e-07,
+      "loss": 0.9901,
+      "step": 3076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.44617823728988e-07,
+      "loss": 0.9901,
+      "step": 3077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.413182821934527e-07,
+      "loss": 1.0074,
+      "step": 3078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.380269270847295e-07,
+      "loss": 0.9337,
+      "step": 3079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.347437612819051e-07,
+      "loss": 0.9457,
+      "step": 3080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.31468787656897e-07,
+      "loss": 0.9858,
+      "step": 3081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.282020090744623e-07,
+      "loss": 0.9468,
+      "step": 3082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.249434283921885e-07,
+      "loss": 1.0227,
+      "step": 3083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.216930484604888e-07,
+      "loss": 0.9854,
+      "step": 3084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.18450872122609e-07,
+      "loss": 0.368,
+      "step": 3085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.152169022146115e-07,
+      "loss": 0.9982,
+      "step": 3086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.119911415653845e-07,
+      "loss": 0.9766,
+      "step": 3087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.087735929966354e-07,
+      "loss": 1.0007,
+      "step": 3088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.055642593228894e-07,
+      "loss": 0.9724,
+      "step": 3089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.023631433514787e-07,
+      "loss": 1.0198,
+      "step": 3090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.991702478825556e-07,
+      "loss": 0.9929,
+      "step": 3091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.959855757090771e-07,
+      "loss": 0.3762,
+      "step": 3092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.928091296168082e-07,
+      "loss": 0.944,
+      "step": 3093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.896409123843194e-07,
+      "loss": 0.9715,
+      "step": 3094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.864809267829785e-07,
+      "loss": 1.0004,
+      "step": 3095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833291755769599e-07,
+      "loss": 1.0223,
+      "step": 3096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.801856615232249e-07,
+      "loss": 0.9924,
+      "step": 3097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.770503873715427e-07,
+      "loss": 1.0003,
+      "step": 3098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.73923355864463e-07,
+      "loss": 0.9902,
+      "step": 3099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.708045697373299e-07,
+      "loss": 0.9949,
+      "step": 3100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.676940317182789e-07,
+      "loss": 0.9982,
+      "step": 3101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.645917445282201e-07,
+      "loss": 0.9738,
+      "step": 3102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.614977108808595e-07,
+      "loss": 1.0252,
+      "step": 3103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.584119334826709e-07,
+      "loss": 0.9971,
+      "step": 3104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.553344150329166e-07,
+      "loss": 1.022,
+      "step": 3105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.522651582236238e-07,
+      "loss": 1.0109,
+      "step": 3106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.492041657396052e-07,
+      "loss": 0.9825,
+      "step": 3107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.461514402584333e-07,
+      "loss": 0.971,
+      "step": 3108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.431069844504544e-07,
+      "loss": 0.9928,
+      "step": 3109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.400708009787826e-07,
+      "loss": 0.9764,
+      "step": 3110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.370428924992887e-07,
+      "loss": 0.9659,
+      "step": 3111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.340232616606144e-07,
+      "loss": 0.9926,
+      "step": 3112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.310119111041534e-07,
+      "loss": 0.9656,
+      "step": 3113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280088434640607e-07,
+      "loss": 1.0036,
+      "step": 3114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.250140613672416e-07,
+      "loss": 1.0088,
+      "step": 3115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.220275674333575e-07,
+      "loss": 0.9766,
+      "step": 3116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.190493642748184e-07,
+      "loss": 0.9503,
+      "step": 3117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.160794544967818e-07,
+      "loss": 1.0194,
+      "step": 3118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.131178406971526e-07,
+      "loss": 1.0231,
+      "step": 3119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.101645254665744e-07,
+      "loss": 1.0094,
+      "step": 3120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.072195113884359e-07,
+      "loss": 1.0272,
+      "step": 3121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.042828010388623e-07,
+      "loss": 0.3456,
+      "step": 3122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.013543969867185e-07,
+      "loss": 0.9572,
+      "step": 3123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984343017935967e-07,
+      "loss": 0.977,
+      "step": 3124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.955225180138268e-07,
+      "loss": 0.9845,
+      "step": 3125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.926190481944682e-07,
+      "loss": 0.9582,
+      "step": 3126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.897238948753047e-07,
+      "loss": 0.9998,
+      "step": 3127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.868370605888495e-07,
+      "loss": 1.0101,
+      "step": 3128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.839585478603326e-07,
+      "loss": 1.0276,
+      "step": 3129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.810883592077098e-07,
+      "loss": 1.0064,
+      "step": 3130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.782264971416539e-07,
+      "loss": 0.9666,
+      "step": 3131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.753729641655569e-07,
+      "loss": 0.9945,
+      "step": 3132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.725277627755187e-07,
+      "loss": 0.9506,
+      "step": 3133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.696908954603552e-07,
+      "loss": 0.929,
+      "step": 3134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.668623647015946e-07,
+      "loss": 0.9873,
+      "step": 3135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6404217297346676e-07,
+      "loss": 0.9633,
+      "step": 3136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6123032274291493e-07,
+      "loss": 0.9802,
+      "step": 3137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.58426816469576e-07,
+      "loss": 0.9951,
+      "step": 3138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5563165660579387e-07,
+      "loss": 0.952,
+      "step": 3139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.528448455966139e-07,
+      "loss": 0.3252,
+      "step": 3140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.500663858797738e-07,
+      "loss": 0.9499,
+      "step": 3141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.47296279885705e-07,
+      "loss": 1.0103,
+      "step": 3142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.44534530037537e-07,
+      "loss": 0.9777,
+      "step": 3143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.417811387510851e-07,
+      "loss": 0.9884,
+      "step": 3144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.39036108434856e-07,
+      "loss": 0.9717,
+      "step": 3145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3629944149004233e-07,
+      "loss": 0.9635,
+      "step": 3146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.335711403105169e-07,
+      "loss": 0.3401,
+      "step": 3147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.308512072828419e-07,
+      "loss": 1.0138,
+      "step": 3148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.281396447862496e-07,
+      "loss": 1.0105,
+      "step": 3149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2543645519266377e-07,
+      "loss": 0.9702,
+      "step": 3150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2274164086666956e-07,
+      "loss": 0.9631,
+      "step": 3151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.20055204165537e-07,
+      "loss": 0.9561,
+      "step": 3152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1737714743920296e-07,
+      "loss": 0.9793,
+      "step": 3153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1470747303027135e-07,
+      "loss": 0.9735,
+      "step": 3154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.12046183274023e-07,
+      "loss": 0.9766,
+      "step": 3155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0939328049839247e-07,
+      "loss": 0.9504,
+      "step": 3156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.067487670239889e-07,
+      "loss": 0.9842,
+      "step": 3157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.04112645164072e-07,
+      "loss": 0.9638,
+      "step": 3158
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0148491722457275e-07,
+      "loss": 0.9516,
+      "step": 3159
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9886558550406904e-07,
+      "loss": 0.9895,
+      "step": 3160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9625465229380265e-07,
+      "loss": 0.9691,
+      "step": 3161
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.936521198776633e-07,
+      "loss": 0.3682,
+      "step": 3162
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9105799053219204e-07,
+      "loss": 1.0172,
+      "step": 3163
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.88472266526585e-07,
+      "loss": 0.9621,
+      "step": 3164
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.858949501226794e-07,
+      "loss": 0.9749,
+      "step": 3165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.833260435749642e-07,
+      "loss": 0.9863,
+      "step": 3166
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8076554913056286e-07,
+      "loss": 0.9826,
+      "step": 3167
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.782134690292505e-07,
+      "loss": 0.9677,
+      "step": 3168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.756698055034358e-07,
+      "loss": 1.0054,
+      "step": 3169
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7313456077816666e-07,
+      "loss": 0.9848,
+      "step": 3170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7060773707112806e-07,
+      "loss": 0.9681,
+      "step": 3171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.680893365926363e-07,
+      "loss": 0.9841,
+      "step": 3172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6557936154563933e-07,
+      "loss": 1.0116,
+      "step": 3173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6307781412571854e-07,
+      "loss": 1.0007,
+      "step": 3174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.605846965210813e-07,
+      "loss": 0.993,
+      "step": 3175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5810001091255766e-07,
+      "loss": 1.0232,
+      "step": 3176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.556237594736089e-07,
+      "loss": 0.9841,
+      "step": 3177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.531559443703125e-07,
+      "loss": 0.9893,
+      "step": 3178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.506965677613705e-07,
+      "loss": 1.0132,
+      "step": 3179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482456317981009e-07,
+      "loss": 1.0317,
+      "step": 3180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.458031386244365e-07,
+      "loss": 0.9898,
+      "step": 3181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4336909037693047e-07,
+      "loss": 1.0114,
+      "step": 3182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4094348918474406e-07,
+      "loss": 0.9971,
+      "step": 3183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3852633716965345e-07,
+      "loss": 0.9453,
+      "step": 3184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3611763644603836e-07,
+      "loss": 1.0074,
+      "step": 3185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3371738912089114e-07,
+      "loss": 0.986,
+      "step": 3186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.313255972938068e-07,
+      "loss": 1.0092,
+      "step": 3187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.289422630569861e-07,
+      "loss": 1.0254,
+      "step": 3188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.265673884952292e-07,
+      "loss": 0.9396,
+      "step": 3189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2420097568593765e-07,
+      "loss": 0.37,
+      "step": 3190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.218430266991102e-07,
+      "loss": 0.9841,
+      "step": 3191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.194935435973434e-07,
+      "loss": 0.9835,
+      "step": 3192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.171525284358279e-07,
+      "loss": 0.9444,
+      "step": 3193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.148199832623444e-07,
+      "loss": 0.3541,
+      "step": 3194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1249591011726865e-07,
+      "loss": 0.9928,
+      "step": 3195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.101803110335644e-07,
+      "loss": 0.9978,
+      "step": 3196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.078731880367802e-07,
+      "loss": 0.9525,
+      "step": 3197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.055745431450552e-07,
+      "loss": 0.9901,
+      "step": 3198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.032843783691053e-07,
+      "loss": 0.9845,
+      "step": 3199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0100269571223696e-07,
+      "loss": 0.3671,
+      "step": 3200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9872949717032695e-07,
+      "loss": 1.0116,
+      "step": 3201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.964647847318425e-07,
+      "loss": 0.9922,
+      "step": 3202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9420856037781667e-07,
+      "loss": 0.9805,
+      "step": 3203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9196082608186516e-07,
+      "loss": 1.055,
+      "step": 3204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.897215838101741e-07,
+      "loss": 0.9847,
+      "step": 3205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.874908355214989e-07,
+      "loss": 0.344,
+      "step": 3206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8526858316717197e-07,
+      "loss": 1.0217,
+      "step": 3207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8305482869108505e-07,
+      "loss": 0.9788,
+      "step": 3208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8084957402970457e-07,
+      "loss": 1.02,
+      "step": 3209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.786528211120554e-07,
+      "loss": 0.9622,
+      "step": 3210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.764645718597303e-07,
+      "loss": 1.0197,
+      "step": 3211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.742848281868804e-07,
+      "loss": 1.0303,
+      "step": 3212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7211359200021824e-07,
+      "loss": 0.9976,
+      "step": 3213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.699508651990157e-07,
+      "loss": 0.9686,
+      "step": 3214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6779664967509613e-07,
+      "loss": 0.9941,
+      "step": 3215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6565094731284436e-07,
+      "loss": 0.3555,
+      "step": 3216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.635137599891924e-07,
+      "loss": 0.9898,
+      "step": 3217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.613850895736303e-07,
+      "loss": 0.9702,
+      "step": 3218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5926493792819085e-07,
+      "loss": 0.9667,
+      "step": 3219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5715330690745946e-07,
+      "loss": 0.3595,
+      "step": 3220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.550501983585674e-07,
+      "loss": 0.9552,
+      "step": 3221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5295561412119197e-07,
+      "loss": 0.9878,
+      "step": 3222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.50869556027552e-07,
+      "loss": 0.9997,
+      "step": 3223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.487920259024079e-07,
+      "loss": 1.0084,
+      "step": 3224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.467230255630637e-07,
+      "loss": 1.0209,
+      "step": 3225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4466255681935723e-07,
+      "loss": 1.0212,
+      "step": 3226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.42610621473669e-07,
+      "loss": 0.9676,
+      "step": 3227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.405672213209087e-07,
+      "loss": 0.9875,
+      "step": 3228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.385323581485244e-07,
+      "loss": 1.0223,
+      "step": 3229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.365060337364955e-07,
+      "loss": 1.0042,
+      "step": 3230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3448824985733197e-07,
+      "loss": 0.9662,
+      "step": 3231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3247900827607417e-07,
+      "loss": 0.9915,
+      "step": 3232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3047831075028616e-07,
+      "loss": 1.0003,
+      "step": 3233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2848615903006355e-07,
+      "loss": 1.0072,
+      "step": 3234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2650255485802243e-07,
+      "loss": 1.005,
+      "step": 3235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245274999693048e-07,
+      "loss": 1.0033,
+      "step": 3236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.225609960915709e-07,
+      "loss": 0.9819,
+      "step": 3237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2060304494500584e-07,
+      "loss": 1.0077,
+      "step": 3238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.186536482423085e-07,
+      "loss": 0.9463,
+      "step": 3239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1671280768869928e-07,
+      "loss": 1.0174,
+      "step": 3240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1478052498191015e-07,
+      "loss": 1.0421,
+      "step": 3241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1285680181218903e-07,
+      "loss": 0.9635,
+      "step": 3242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1094163986229764e-07,
+      "loss": 0.9625,
+      "step": 3243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.090350408075048e-07,
+      "loss": 1.0187,
+      "step": 3244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0713700631559419e-07,
+      "loss": 0.9875,
+      "step": 3245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0524753804685437e-07,
+      "loss": 0.9903,
+      "step": 3246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0336663765408104e-07,
+      "loss": 0.9953,
+      "step": 3247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0149430678257698e-07,
+      "loss": 1.0187,
+      "step": 3248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.996305470701465e-07,
+      "loss": 0.9787,
+      "step": 3249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9777536014709886e-07,
+      "loss": 0.9712,
+      "step": 3250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9592874763624148e-07,
+      "loss": 1.0201,
+      "step": 3251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9409071115288447e-07,
+      "loss": 1.0243,
+      "step": 3252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.922612523048306e-07,
+      "loss": 0.9894,
+      "step": 3253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.904403726923887e-07,
+      "loss": 1.003,
+      "step": 3254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8862807390835457e-07,
+      "loss": 0.9735,
+      "step": 3255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.868243575380213e-07,
+      "loss": 0.9339,
+      "step": 3256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8502922515917565e-07,
+      "loss": 0.9942,
+      "step": 3257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8324267834209153e-07,
+      "loss": 0.9469,
+      "step": 3258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8146471864953885e-07,
+      "loss": 0.9647,
+      "step": 3259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7969534763677132e-07,
+      "loss": 1.0292,
+      "step": 3260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7793456685153088e-07,
+      "loss": 0.9853,
+      "step": 3261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7618237783404436e-07,
+      "loss": 0.9874,
+      "step": 3262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.744387821170268e-07,
+      "loss": 0.3438,
+      "step": 3263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7270378122567044e-07,
+      "loss": 0.9683,
+      "step": 3264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7097737667765568e-07,
+      "loss": 0.9728,
+      "step": 3265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6925956998313787e-07,
+      "loss": 0.9421,
+      "step": 3266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.675503626447539e-07,
+      "loss": 0.9797,
+      "step": 3267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6584975615762e-07,
+      "loss": 0.9613,
+      "step": 3268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6415775200932626e-07,
+      "loss": 1.0011,
+      "step": 3269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6247435167993874e-07,
+      "loss": 0.9861,
+      "step": 3270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607995566419962e-07,
+      "loss": 0.9471,
+      "step": 3271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.591333683605123e-07,
+      "loss": 0.9862,
+      "step": 3272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5747578829297228e-07,
+      "loss": 0.9568,
+      "step": 3273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.558268178893274e-07,
+      "loss": 0.3506,
+      "step": 3274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5418645859200387e-07,
+      "loss": 0.9772,
+      "step": 3275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5255471183588942e-07,
+      "loss": 1.0207,
+      "step": 3276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5093157904834012e-07,
+      "loss": 0.9918,
+      "step": 3277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4931706164917902e-07,
+      "loss": 1.0279,
+      "step": 3278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4771116105069206e-07,
+      "loss": 1.0208,
+      "step": 3279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.461138786576255e-07,
+      "loss": 0.9888,
+      "step": 3280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4452521586718838e-07,
+      "loss": 1.039,
+      "step": 3281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4294517406905127e-07,
+      "loss": 0.9789,
+      "step": 3282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4137375464534086e-07,
+      "loss": 1.0225,
+      "step": 3283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3981095897064424e-07,
+      "loss": 1.0,
+      "step": 3284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3825678841200118e-07,
+      "loss": 1.0151,
+      "step": 3285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3671124432891202e-07,
+      "loss": 1.0291,
+      "step": 3286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3517432807332641e-07,
+      "loss": 1.0228,
+      "step": 3287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3364604098964896e-07,
+      "loss": 0.9746,
+      "step": 3288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.321263844147358e-07,
+      "loss": 1.0212,
+      "step": 3289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3061535967789364e-07,
+      "loss": 0.9582,
+      "step": 3290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.291129681008785e-07,
+      "loss": 1.0394,
+      "step": 3291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2761921099789464e-07,
+      "loss": 1.0001,
+      "step": 3292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.261340896755947e-07,
+      "loss": 1.0075,
+      "step": 3293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.246576054330728e-07,
+      "loss": 0.962,
+      "step": 3294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.231897595618725e-07,
+      "loss": 0.9899,
+      "step": 3295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2173055334597893e-07,
+      "loss": 0.9922,
+      "step": 3296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2027998806182107e-07,
+      "loss": 1.0345,
+      "step": 3297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1883806497826833e-07,
+      "loss": 1.0069,
+      "step": 3298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1740478535663069e-07,
+      "loss": 0.9815,
+      "step": 3299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1598015045065636e-07,
+      "loss": 1.0001,
+      "step": 3300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1456416150653294e-07,
+      "loss": 0.9489,
+      "step": 3301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1315681976288629e-07,
+      "loss": 0.371,
+      "step": 3302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1175812645077388e-07,
+      "loss": 0.9696,
+      "step": 3303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.103680827936926e-07,
+      "loss": 0.9879,
+      "step": 3304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0898669000756978e-07,
+      "loss": 0.9571,
+      "step": 3305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0761394930076885e-07,
+      "loss": 0.9571,
+      "step": 3306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0624986187408038e-07,
+      "loss": 0.3628,
+      "step": 3307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0489442892073098e-07,
+      "loss": 0.9137,
+      "step": 3308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0354765162637337e-07,
+      "loss": 0.9431,
+      "step": 3309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0220953116908738e-07,
+      "loss": 0.9912,
+      "step": 3310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.008800687193845e-07,
+      "loss": 1.0231,
+      "step": 3311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.95592654402e-08,
+      "loss": 1.0285,
+      "step": 3312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.824712248689639e-08,
+      "loss": 1.0093,
+      "step": 3313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.694364100725662e-08,
+      "loss": 0.9835,
+      "step": 3314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.564882214149418e-08,
+      "loss": 1.0168,
+      "step": 3315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.43626670222364e-08,
+      "loss": 0.9336,
+      "step": 3316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.308517677453887e-08,
+      "loss": 1.001,
+      "step": 3317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.181635251587661e-08,
+      "loss": 0.943,
+      "step": 3318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.055619535613847e-08,
+      "loss": 0.9582,
+      "step": 3319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.930470639764043e-08,
+      "loss": 1.0019,
+      "step": 3320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.806188673511018e-08,
+      "loss": 0.9917,
+      "step": 3321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.682773745569361e-08,
+      "loss": 0.9453,
+      "step": 3322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.560225963895163e-08,
+      "loss": 1.0107,
+      "step": 3323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.438545435686008e-08,
+      "loss": 1.0096,
+      "step": 3324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.317732267381084e-08,
+      "loss": 0.992,
+      "step": 3325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.197786564660636e-08,
+      "loss": 0.9589,
+      "step": 3326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.078708432446069e-08,
+      "loss": 1.0235,
+      "step": 3327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.960497974899839e-08,
+      "loss": 0.9571,
+      "step": 3328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.843155295425786e-08,
+      "loss": 0.9912,
+      "step": 3329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.72668049666836e-08,
+      "loss": 0.9792,
+      "step": 3330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.611073680512726e-08,
+      "loss": 0.9498,
+      "step": 3331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.496334948085104e-08,
+      "loss": 0.9823,
+      "step": 3332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.382464399752209e-08,
+      "loss": 0.9924,
+      "step": 3333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.269462135121364e-08,
+      "loss": 0.9536,
+      "step": 3334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.157328253040163e-08,
+      "loss": 0.9682,
+      "step": 3335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.046062851596924e-08,
+      "loss": 0.9987,
+      "step": 3336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.935666028120125e-08,
+      "loss": 1.0183,
+      "step": 3337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.826137879178296e-08,
+      "loss": 0.9452,
+      "step": 3338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.717478500580243e-08,
+      "loss": 0.9926,
+      "step": 3339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.609687987375047e-08,
+      "loss": 0.9857,
+      "step": 3340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.502766433851393e-08,
+      "loss": 1.0072,
+      "step": 3341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.396713933537913e-08,
+      "loss": 1.0132,
+      "step": 3342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.291530579203286e-08,
+      "loss": 0.3578,
+      "step": 3343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.187216462855694e-08,
+      "loss": 1.0218,
+      "step": 3344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.083771675743033e-08,
+      "loss": 0.9867,
+      "step": 3345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9811963083527e-08,
+      "loss": 0.9905,
+      "step": 3346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.879490450411585e-08,
+      "loss": 1.013,
+      "step": 3347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7786541908860796e-08,
+      "loss": 1.0,
+      "step": 3348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.678687617981848e-08,
+      "loss": 1.018,
+      "step": 3349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.579590819143832e-08,
+      "loss": 0.96,
+      "step": 3350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.481363881056023e-08,
+      "loss": 1.0254,
+      "step": 3351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3840068896415795e-08,
+      "loss": 0.9828,
+      "step": 3352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2875199300628235e-08,
+      "loss": 0.9534,
+      "step": 3353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1919030867207955e-08,
+      "loss": 0.9994,
+      "step": 3354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.097156443255591e-08,
+      "loss": 0.9812,
+      "step": 3355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0032800825459135e-08,
+      "loss": 0.9988,
+      "step": 3356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.910274086709299e-08,
+      "loss": 0.9921,
+      "step": 3357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.818138537102113e-08,
+      "loss": 0.9634,
+      "step": 3358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.726873514318997e-08,
+      "loss": 0.9742,
+      "step": 3359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.636479098193314e-08,
+      "loss": 0.9997,
+      "step": 3360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.546955367796924e-08,
+      "loss": 0.9869,
+      "step": 3361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.458302401439518e-08,
+      "loss": 0.974,
+      "step": 3362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.370520276670065e-08,
+      "loss": 1.0092,
+      "step": 3363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.283609070274808e-08,
+      "loss": 0.9966,
+      "step": 3364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.197568858278933e-08,
+      "loss": 0.9391,
+      "step": 3365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.112399715945015e-08,
+      "loss": 0.982,
+      "step": 3366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.028101717774236e-08,
+      "loss": 0.3435,
+      "step": 3367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.94467493750561e-08,
+      "loss": 0.9837,
+      "step": 3368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8621194481158754e-08,
+      "loss": 0.9848,
+      "step": 3369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7804353218197086e-08,
+      "loss": 0.9694,
+      "step": 3370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.699622630069733e-08,
+      "loss": 1.0526,
+      "step": 3371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6196814435561775e-08,
+      "loss": 0.9847,
+      "step": 3372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.540611832206664e-08,
+      "loss": 1.0294,
+      "step": 3373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.462413865186976e-08,
+      "loss": 0.9907,
+      "step": 3374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385087610899951e-08,
+      "loss": 0.9681,
+      "step": 3375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.308633136986039e-08,
+      "loss": 1.028,
+      "step": 3376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.233050510323077e-08,
+      "loss": 0.9818,
+      "step": 3377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.158339797026622e-08,
+      "loss": 0.9836,
+      "step": 3378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0845010624489525e-08,
+      "loss": 1.0248,
+      "step": 3379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.011534371179958e-08,
+      "loss": 0.9895,
+      "step": 3380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.939439787046583e-08,
+      "loss": 0.9615,
+      "step": 3381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8682173731130468e-08,
+      "loss": 1.0013,
+      "step": 3382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7978671916805145e-08,
+      "loss": 0.9422,
+      "step": 3383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7283893042870935e-08,
+      "loss": 0.9742,
+      "step": 3384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6597837717081687e-08,
+      "loss": 1.0123,
+      "step": 3385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5920506539556244e-08,
+      "loss": 0.9688,
+      "step": 3386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.525190010278622e-08,
+      "loss": 0.9624,
+      "step": 3387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.459201899162822e-08,
+      "loss": 1.0113,
+      "step": 3388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3940863783308288e-08,
+      "loss": 1.0273,
+      "step": 3389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3298435047417468e-08,
+      "loss": 0.9895,
+      "step": 3390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2664733345916234e-08,
+      "loss": 1.0277,
+      "step": 3391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2039759233130066e-08,
+      "loss": 0.9546,
+      "step": 3392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1423513255748318e-08,
+      "loss": 0.9978,
+      "step": 3393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0815995952828684e-08,
+      "loss": 0.9274,
+      "step": 3394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02172078557894e-08,
+      "loss": 0.9976,
+      "step": 3395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9627149488415932e-08,
+      "loss": 0.9941,
+      "step": 3396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.904582136685873e-08,
+      "loss": 0.9889,
+      "step": 3397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8473223999627697e-08,
+      "loss": 0.9839,
+      "step": 3398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7909357887597732e-08,
+      "loss": 0.983,
+      "step": 3399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.735422352400651e-08,
+      "loss": 1.0194,
+      "step": 3400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6807821394452252e-08,
+      "loss": 1.0348,
+      "step": 3401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6270151976895964e-08,
+      "loss": 0.9672,
+      "step": 3402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574121574165921e-08,
+      "loss": 0.9878,
+      "step": 3403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5221013151425214e-08,
+      "loss": 0.9817,
+      "step": 3404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.470954466123553e-08,
+      "loss": 1.0044,
+      "step": 3405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4206810718494501e-08,
+      "loss": 1.0179,
+      "step": 3406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3712811762962574e-08,
+      "loss": 1.0055,
+      "step": 3407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3227548226761866e-08,
+      "loss": 0.9347,
+      "step": 3408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2751020534371716e-08,
+      "loss": 1.0219,
+      "step": 3409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2283229102633131e-08,
+      "loss": 1.0474,
+      "step": 3410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1824174340739902e-08,
+      "loss": 0.9386,
+      "step": 3411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1373856650247484e-08,
+      "loss": 1.04,
+      "step": 3412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0932276425067446e-08,
+      "loss": 0.9526,
+      "step": 3413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0499434051467472e-08,
+      "loss": 0.9861,
+      "step": 3414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0075329908073584e-08,
+      "loss": 1.013,
+      "step": 3415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.659964365866803e-09,
+      "loss": 0.9923,
+      "step": 3416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.25333778818538e-09,
+      "loss": 0.9545,
+      "step": 3417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.855450530721455e-09,
+      "loss": 1.0442,
+      "step": 3418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.466302941524396e-09,
+      "loss": 0.9241,
+      "step": 3419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.085895360997465e-09,
+      "loss": 0.9778,
+      "step": 3420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.714228121897815e-09,
+      "loss": 0.9859,
+      "step": 3421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.351301549339829e-09,
+      "loss": 1.0054,
+      "step": 3422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.997115960789558e-09,
+      "loss": 1.0501,
+      "step": 3423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.6516716660680605e-09,
+      "loss": 0.9436,
+      "step": 3424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.314968967349178e-09,
+      "loss": 0.9854,
+      "step": 3425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.987008159161756e-09,
+      "loss": 1.0029,
+      "step": 3426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.667789528386314e-09,
+      "loss": 0.9847,
+      "step": 3427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.357313354256155e-09,
+      "loss": 0.943,
+      "step": 3428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.055579908357366e-09,
+      "loss": 0.9865,
+      "step": 3429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.762589454629929e-09,
+      "loss": 0.9955,
+      "step": 3430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4783422493643865e-09,
+      "loss": 1.0343,
+      "step": 3431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2028385412029585e-09,
+      "loss": 0.9744,
+      "step": 3432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.936078571140644e-09,
+      "loss": 0.3538,
+      "step": 3433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6780625725230114e-09,
+      "loss": 0.3588,
+      "step": 3434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4287907710484068e-09,
+      "loss": 1.0031,
+      "step": 3435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1882633847635235e-09,
+      "loss": 1.0166,
+      "step": 3436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.956480624068947e-09,
+      "loss": 1.0129,
+      "step": 3437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7334426917136058e-09,
+      "loss": 1.0433,
+      "step": 3438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.519149782799213e-09,
+      "loss": 0.9562,
+      "step": 3439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3136020847769334e-09,
+      "loss": 1.0195,
+      "step": 3440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1167997774462765e-09,
+      "loss": 1.0148,
+      "step": 3441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9287430329584244e-09,
+      "loss": 0.9896,
+      "step": 3442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7494320158162325e-09,
+      "loss": 0.9863,
+      "step": 3443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5788668828686792e-09,
+      "loss": 0.9981,
+      "step": 3444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4170477833175268e-09,
+      "loss": 1.0304,
+      "step": 3445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2639748587117695e-09,
+      "loss": 0.9915,
+      "step": 3446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1196482429509658e-09,
+      "loss": 0.3374,
+      "step": 3447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.840680622830168e-10,
+      "loss": 1.0485,
+      "step": 3448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.572344353063866e-10,
+      "loss": 0.3503,
+      "step": 3449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.391474729667725e-10,
+      "loss": 1.0132,
+      "step": 3450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.298072785604347e-10,
+      "loss": 0.9665,
+      "step": 3451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.292139477319768e-10,
+      "loss": 0.3546,
+      "step": 3452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.373675684732348e-10,
+      "loss": 0.9611,
+      "step": 3453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5426822112771817e-10,
+      "loss": 0.9994,
+      "step": 3454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7991597838505913e-10,
+      "loss": 1.004,
+      "step": 3455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1431090528434285e-10,
+      "loss": 0.3511,
+      "step": 3456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5745305921299748e-10,
+      "loss": 1.0406,
+      "step": 3457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.093424899079043e-10,
+      "loss": 0.975,
+      "step": 3458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.997923945095685e-11,
+      "loss": 0.978,
+      "step": 3459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9363342276832475e-11,
+      "loss": 0.9926,
+      "step": 3460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749482516633094e-11,
+      "loss": 0.9533,
+      "step": 3461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.37370724748476e-12,
+      "loss": 0.9568,
+      "step": 3462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.9997,
+      "step": 3463
+    },
+    {
+      "epoch": 1.0,
+      "step": 3463,
+      "total_flos": 1.9220823824538796e+19,
+      "train_loss": 1.018446381918929,
+      "train_runtime": 49305.2535,
+      "train_samples_per_second": 13.485,
+      "train_steps_per_second": 0.07
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3463,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 1.9220823824538796e+19,
+  "train_batch_size": 12,
+  "trial_name": null,
+  "trial_params": null
+}