diff --git "a/Chinese_tiny_llm-zh_105k_en_13k/trainer_state.json" "b/Chinese_tiny_llm-zh_105k_en_13k/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/Chinese_tiny_llm-zh_105k_en_13k/trainer_state.json"
@@ -0,0 +1,16653 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2772,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4388489208633095e-07,
+      "loss": 2.3547,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.877697841726619e-07,
+      "loss": 2.3576,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3165467625899287e-07,
+      "loss": 2.1275,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.755395683453238e-07,
+      "loss": 2.4731,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.194244604316547e-07,
+      "loss": 2.4011,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.633093525179857e-07,
+      "loss": 2.3406,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0071942446043167e-06,
+      "loss": 2.4751,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1510791366906476e-06,
+      "loss": 2.145,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2949640287769785e-06,
+      "loss": 2.7092,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4388489208633094e-06,
+      "loss": 2.3121,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5827338129496403e-06,
+      "loss": 2.6237,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7266187050359715e-06,
+      "loss": 2.3887,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8705035971223024e-06,
+      "loss": 2.3881,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0143884892086333e-06,
+      "loss": 2.1992,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.158273381294964e-06,
+      "loss": 2.6365,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.302158273381295e-06,
+      "loss": 2.1901,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4460431654676263e-06,
+      "loss": 2.2422,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.589928057553957e-06,
+      "loss": 2.2557,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.733812949640288e-06,
+      "loss": 2.3875,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.877697841726619e-06,
+      "loss": 2.3422,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.02158273381295e-06,
+      "loss": 2.1384,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1654676258992807e-06,
+      "loss": 1.8947,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.309352517985612e-06,
+      "loss": 2.0175,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.453237410071943e-06,
+      "loss": 2.0965,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5971223021582737e-06,
+      "loss": 2.1806,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.741007194244605e-06,
+      "loss": 2.0644,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.884892086330936e-06,
+      "loss": 1.9973,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.028776978417267e-06,
+      "loss": 1.8138,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.172661870503597e-06,
+      "loss": 2.0755,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.316546762589928e-06,
+      "loss": 1.8519,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.46043165467626e-06,
+      "loss": 2.2144,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.60431654676259e-06,
+      "loss": 1.9688,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.748201438848921e-06,
+      "loss": 2.0193,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.892086330935253e-06,
+      "loss": 1.8964,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.035971223021583e-06,
+      "loss": 2.0533,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.179856115107914e-06,
+      "loss": 2.1172,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.3237410071942456e-06,
+      "loss": 2.1439,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.467625899280576e-06,
+      "loss": 1.8953,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.611510791366906e-06,
+      "loss": 2.0619,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.755395683453238e-06,
+      "loss": 2.0246,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.899280575539568e-06,
+      "loss": 1.9229,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0431654676259e-06,
+      "loss": 1.9145,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.1870503597122315e-06,
+      "loss": 2.159,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.330935251798561e-06,
+      "loss": 1.91,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.474820143884892e-06,
+      "loss": 1.9358,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.618705035971224e-06,
+      "loss": 1.9698,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.762589928057554e-06,
+      "loss": 1.673,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.906474820143886e-06,
+      "loss": 1.9999,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.050359712230216e-06,
+      "loss": 2.0436,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.194244604316547e-06,
+      "loss": 2.0187,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.338129496402878e-06,
+      "loss": 1.8008,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.48201438848921e-06,
+      "loss": 2.1516,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.62589928057554e-06,
+      "loss": 1.8388,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.769784172661872e-06,
+      "loss": 1.9828,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.913669064748202e-06,
+      "loss": 1.7816,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.057553956834533e-06,
+      "loss": 1.8797,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.201438848920865e-06,
+      "loss": 1.8773,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.345323741007195e-06,
+      "loss": 1.8915,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.489208633093526e-06,
+      "loss": 1.8354,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.633093525179856e-06,
+      "loss": 1.9031,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.776978417266188e-06,
+      "loss": 1.8754,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.92086330935252e-06,
+      "loss": 1.772,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.064748201438849e-06,
+      "loss": 1.8173,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.20863309352518e-06,
+      "loss": 1.8111,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.35251798561151e-06,
+      "loss": 1.8438,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.496402877697842e-06,
+      "loss": 1.5675,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.640287769784174e-06,
+      "loss": 1.9983,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784172661870505e-06,
+      "loss": 1.8764,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.928057553956835e-06,
+      "loss": 1.9815,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0071942446043167e-05,
+      "loss": 1.9521,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0215827338129498e-05,
+      "loss": 1.7376,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0359712230215828e-05,
+      "loss": 1.9935,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0503597122302158e-05,
+      "loss": 2.1314,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0647482014388491e-05,
+      "loss": 1.8567,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0791366906474821e-05,
+      "loss": 2.1048,
+      "step": 75
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0935251798561153e-05,
+      "loss": 2.0353,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1079136690647482e-05,
+      "loss": 1.8824,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1223021582733812e-05,
+      "loss": 2.0055,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1366906474820146e-05,
+      "loss": 1.8171,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1510791366906475e-05,
+      "loss": 1.9234,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1654676258992807e-05,
+      "loss": 1.8208,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1798561151079137e-05,
+      "loss": 1.7002,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1942446043165468e-05,
+      "loss": 1.907,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.20863309352518e-05,
+      "loss": 1.9767,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.223021582733813e-05,
+      "loss": 1.7917,
+      "step": 85
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2374100719424463e-05,
+      "loss": 1.6337,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2517985611510793e-05,
+      "loss": 2.0116,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2661870503597123e-05,
+      "loss": 2.0588,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2805755395683454e-05,
+      "loss": 1.5452,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2949640287769784e-05,
+      "loss": 1.8182,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3093525179856117e-05,
+      "loss": 2.0173,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3237410071942447e-05,
+      "loss": 1.5917,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3381294964028777e-05,
+      "loss": 1.845,
+      "step": 93
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3525179856115109e-05,
+      "loss": 1.9642,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3669064748201439e-05,
+      "loss": 2.0804,
+      "step": 95
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3812949640287772e-05,
+      "loss": 1.8469,
+      "step": 96
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3956834532374102e-05,
+      "loss": 2.0403,
+      "step": 97
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4100719424460432e-05,
+      "loss": 2.1303,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4244604316546765e-05,
+      "loss": 2.1525,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4388489208633095e-05,
+      "loss": 1.6616,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4532374100719426e-05,
+      "loss": 2.0293,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4676258992805756e-05,
+      "loss": 2.0534,
+      "step": 102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4820143884892086e-05,
+      "loss": 2.0292,
+      "step": 103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.496402877697842e-05,
+      "loss": 1.8536,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5107913669064749e-05,
+      "loss": 1.8581,
+      "step": 105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.525179856115108e-05,
+      "loss": 1.6334,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5395683453237412e-05,
+      "loss": 1.8721,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5539568345323744e-05,
+      "loss": 1.8919,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5683453237410072e-05,
+      "loss": 1.7582,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5827338129496403e-05,
+      "loss": 1.7654,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5971223021582735e-05,
+      "loss": 1.9529,
+      "step": 111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6115107913669067e-05,
+      "loss": 2.0011,
+      "step": 112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6258992805755398e-05,
+      "loss": 1.7407,
+      "step": 113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.640287769784173e-05,
+      "loss": 2.0088,
+      "step": 114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6546762589928058e-05,
+      "loss": 2.0356,
+      "step": 115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.669064748201439e-05,
+      "loss": 1.9327,
+      "step": 116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.683453237410072e-05,
+      "loss": 2.0276,
+      "step": 117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6978417266187053e-05,
+      "loss": 1.7289,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7122302158273384e-05,
+      "loss": 2.1943,
+      "step": 119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7266187050359712e-05,
+      "loss": 1.8819,
+      "step": 120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7410071942446044e-05,
+      "loss": 2.142,
+      "step": 121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7553956834532375e-05,
+      "loss": 1.9633,
+      "step": 122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7697841726618707e-05,
+      "loss": 1.9181,
+      "step": 123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.784172661870504e-05,
+      "loss": 1.9825,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7985611510791367e-05,
+      "loss": 1.9971,
+      "step": 125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8129496402877698e-05,
+      "loss": 1.9008,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.827338129496403e-05,
+      "loss": 2.1904,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.841726618705036e-05,
+      "loss": 2.0583,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8561151079136693e-05,
+      "loss": 2.1272,
+      "step": 129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.870503597122302e-05,
+      "loss": 2.0181,
+      "step": 130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8848920863309356e-05,
+      "loss": 1.873,
+      "step": 131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8992805755395684e-05,
+      "loss": 1.8865,
+      "step": 132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9136690647482016e-05,
+      "loss": 1.7558,
+      "step": 133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280575539568347e-05,
+      "loss": 1.9659,
+      "step": 134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9424460431654675e-05,
+      "loss": 1.8793,
+      "step": 135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.956834532374101e-05,
+      "loss": 1.7461,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.971223021582734e-05,
+      "loss": 1.9957,
+      "step": 137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.985611510791367e-05,
+      "loss": 1.8534,
+      "step": 138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 1.8629,
+      "step": 139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999992881836162e-05,
+      "loss": 1.8998,
+      "step": 140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999971527354777e-05,
+      "loss": 1.869,
+      "step": 141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999935936586245e-05,
+      "loss": 1.951,
+      "step": 142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999988610958124e-05,
+      "loss": 1.9328,
+      "step": 143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999982204641069e-05,
+      "loss": 1.6579,
+      "step": 144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999743747165806e-05,
+      "loss": 1.817,
+      "step": 145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999651211958052e-05,
+      "loss": 1.6378,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999544440919166e-05,
+      "loss": 1.863,
+      "step": 147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999423434201147e-05,
+      "loss": 1.8736,
+      "step": 148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999288191976264e-05,
+      "loss": 1.9776,
+      "step": 149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999913871443706e-05,
+      "loss": 1.9897,
+      "step": 150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9998975001796328e-05,
+      "loss": 1.8387,
+      "step": 151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999879705428714e-05,
+      "loss": 1.6932,
+      "step": 152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998604872162825e-05,
+      "loss": 2.2017,
+      "step": 153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998398455696983e-05,
+      "loss": 1.8806,
+      "step": 154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998177805183467e-05,
+      "loss": 1.7912,
+      "step": 155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999794292093641e-05,
+      "loss": 1.6728,
+      "step": 156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99976938032902e-05,
+      "loss": 2.026,
+      "step": 157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999743045259949e-05,
+      "loss": 2.1211,
+      "step": 158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999715286923919e-05,
+      "loss": 1.8448,
+      "step": 159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999686105360448e-05,
+      "loss": 2.0976,
+      "step": 160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9996555006110802e-05,
+      "loss": 1.9568,
+      "step": 161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9996234727193845e-05,
+      "loss": 1.7825,
+      "step": 162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9995900217309575e-05,
+      "loss": 1.7175,
+      "step": 163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999555147693421e-05,
+      "loss": 1.6861,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9995188506564234e-05,
+      "loss": 2.016,
+      "step": 165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994811306716374e-05,
+      "loss": 1.9465,
+      "step": 166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994419877927632e-05,
+      "loss": 2.0717,
+      "step": 167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994014220755255e-05,
+      "loss": 2.1168,
+      "step": 168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999359433577675e-05,
+      "loss": 1.8352,
+      "step": 169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9993160223589875e-05,
+      "loss": 1.8891,
+      "step": 170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9992711884812656e-05,
+      "loss": 1.6747,
+      "step": 171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9992249320083355e-05,
+      "loss": 2.0694,
+      "step": 172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9991772530060497e-05,
+      "loss": 2.0438,
+      "step": 173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999128151542286e-05,
+      "loss": 1.9977,
+      "step": 174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9990776276869456e-05,
+      "loss": 1.6287,
+      "step": 175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9990256815119572e-05,
+      "loss": 1.8438,
+      "step": 176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9989723130912722e-05,
+      "loss": 1.7703,
+      "step": 177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9989175225008685e-05,
+      "loss": 1.8473,
+      "step": 178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9988613098187476e-05,
+      "loss": 1.8733,
+      "step": 179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9988036751249348e-05,
+      "loss": 1.8972,
+      "step": 180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9987446185014815e-05,
+      "loss": 2.0464,
+      "step": 181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9986841400324625e-05,
+      "loss": 1.7597,
+      "step": 182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998622239803977e-05,
+      "loss": 1.8031,
+      "step": 183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9985589179041485e-05,
+      "loss": 2.103,
+      "step": 184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9984941744231233e-05,
+      "loss": 1.9884,
+      "step": 185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998428009453073e-05,
+      "loss": 1.8569,
+      "step": 186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998360423088192e-05,
+      "loss": 2.0007,
+      "step": 187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9982914154246986e-05,
+      "loss": 1.7048,
+      "step": 188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998220986560834e-05,
+      "loss": 2.0295,
+      "step": 189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9981491365968635e-05,
+      "loss": 1.6678,
+      "step": 190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.998075865635075e-05,
+      "loss": 1.801,
+      "step": 191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.998001173779779e-05,
+      "loss": 1.7313,
+      "step": 192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99792506113731e-05,
+      "loss": 1.8665,
+      "step": 193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9978475278160236e-05,
+      "loss": 1.8131,
+      "step": 194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9977685739262996e-05,
+      "loss": 1.9217,
+      "step": 195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9976881995805387e-05,
+      "loss": 1.9435,
+      "step": 196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9976064048931648e-05,
+      "loss": 2.0235,
+      "step": 197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9975231899806233e-05,
+      "loss": 1.8873,
+      "step": 198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997438554961382e-05,
+      "loss": 2.0316,
+      "step": 199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9973524999559295e-05,
+      "loss": 1.7718,
+      "step": 200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9972650250867773e-05,
+      "loss": 2.0169,
+      "step": 201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997176130478457e-05,
+      "loss": 1.9306,
+      "step": 202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997085816257522e-05,
+      "loss": 1.9852,
+      "step": 203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9969940825525462e-05,
+      "loss": 1.8372,
+      "step": 204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9969009294941252e-05,
+      "loss": 1.8715,
+      "step": 205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9968063572148748e-05,
+      "loss": 1.863,
+      "step": 206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9967103658494308e-05,
+      "loss": 1.8807,
+      "step": 207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99661295553445e-05,
+      "loss": 1.9153,
+      "step": 208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9965141264086084e-05,
+      "loss": 2.0213,
+      "step": 209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996413878612603e-05,
+      "loss": 1.707,
+      "step": 210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99631221228915e-05,
+      "loss": 1.8745,
+      "step": 211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9962091275829844e-05,
+      "loss": 1.7234,
+      "step": 212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9961046246408603e-05,
+      "loss": 1.8484,
+      "step": 213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995998703611553e-05,
+      "loss": 1.7754,
+      "step": 214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958913646458546e-05,
+      "loss": 2.0296,
+      "step": 215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9957826078965756e-05,
+      "loss": 1.5189,
+      "step": 216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9956724335185466e-05,
+      "loss": 1.4644,
+      "step": 217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9955608416686147e-05,
+      "loss": 1.9898,
+      "step": 218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995447832505647e-05,
+      "loss": 2.1078,
+      "step": 219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9953334061905254e-05,
+      "loss": 1.7983,
+      "step": 220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995217562886152e-05,
+      "loss": 1.7168,
+      "step": 221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995100302757445e-05,
+      "loss": 1.8447,
+      "step": 222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9949816259713394e-05,
+      "loss": 1.9374,
+      "step": 223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.994861532696788e-05,
+      "loss": 2.0859,
+      "step": 224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9947400231047584e-05,
+      "loss": 1.8303,
+      "step": 225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9946170973682367e-05,
+      "loss": 1.5848,
+      "step": 226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9944927556622236e-05,
+      "loss": 1.8136,
+      "step": 227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9943669981637367e-05,
+      "loss": 1.7803,
+      "step": 228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994239825051807e-05,
+      "loss": 1.9902,
+      "step": 229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9941112365074836e-05,
+      "loss": 1.7615,
+      "step": 230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993981232713829e-05,
+      "loss": 2.0408,
+      "step": 231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9938498138559205e-05,
+      "loss": 1.6415,
+      "step": 232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993716980120851e-05,
+      "loss": 1.8852,
+      "step": 233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9935827316977266e-05,
+      "loss": 1.9254,
+      "step": 234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9934470687776674e-05,
+      "loss": 1.9132,
+      "step": 235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993309991553808e-05,
+      "loss": 1.8581,
+      "step": 236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993171500221296e-05,
+      "loss": 1.8959,
+      "step": 237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993031594977292e-05,
+      "loss": 1.8841,
+      "step": 238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99289027602097e-05,
+      "loss": 1.8922,
+      "step": 239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9927475435535156e-05,
+      "loss": 1.66,
+      "step": 240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9926033977781284e-05,
+      "loss": 2.0959,
+      "step": 241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9924578389000185e-05,
+      "loss": 1.8981,
+      "step": 242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.992310867126408e-05,
+      "loss": 2.079,
+      "step": 243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9921624826665316e-05,
+      "loss": 1.896,
+      "step": 244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9920126857316334e-05,
+      "loss": 2.0524,
+      "step": 245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99186147653497e-05,
+      "loss": 1.7147,
+      "step": 246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991708855291807e-05,
+      "loss": 1.8989,
+      "step": 247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9915548222194215e-05,
+      "loss": 1.6191,
+      "step": 248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9913993775371e-05,
+      "loss": 1.8722,
+      "step": 249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9912425214661386e-05,
+      "loss": 1.8121,
+      "step": 250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910842542298425e-05,
+      "loss": 1.8755,
+      "step": 251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909245760535263e-05,
+      "loss": 2.0044,
+      "step": 252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.990763487164513e-05,
+      "loss": 1.7097,
+      "step": 253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9906009877921345e-05,
+      "loss": 2.0125,
+      "step": 254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9904370781677294e-05,
+      "loss": 1.9329,
+      "step": 255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9902717585246456e-05,
+      "loss": 2.0188,
+      "step": 256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990105029098237e-05,
+      "loss": 1.8302,
+      "step": 257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9899368901258652e-05,
+      "loss": 1.9421,
+      "step": 258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989767341846899e-05,
+      "loss": 1.952,
+      "step": 259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989596384502712e-05,
+      "loss": 1.7265,
+      "step": 260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9894240183366847e-05,
+      "loss": 2.014,
+      "step": 261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989250243594204e-05,
+      "loss": 1.7925,
+      "step": 262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9890750605226606e-05,
+      "loss": 1.9085,
+      "step": 263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.988898469371451e-05,
+      "loss": 1.7916,
+      "step": 264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9887204703919764e-05,
+      "loss": 1.5943,
+      "step": 265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9885410638376418e-05,
+      "loss": 2.0056,
+      "step": 266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9883602499638565e-05,
+      "loss": 1.7821,
+      "step": 267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9881780290280327e-05,
+      "loss": 2.0791,
+      "step": 268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9879944012895863e-05,
+      "loss": 1.8741,
+      "step": 269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9878093670099355e-05,
+      "loss": 1.9063,
+      "step": 270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9876229264525012e-05,
+      "loss": 1.9244,
+      "step": 271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.987435079882707e-05,
+      "loss": 1.9378,
+      "step": 272
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9872458275679766e-05,
+      "loss": 1.8986,
+      "step": 273
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.987055169777736e-05,
+      "loss": 1.8158,
+      "step": 274
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.986863106783412e-05,
+      "loss": 1.4929,
+      "step": 275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9866696388584316e-05,
+      "loss": 1.8164,
+      "step": 276
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9864747662782226e-05,
+      "loss": 1.6363,
+      "step": 277
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9862784893202113e-05,
+      "loss": 1.7578,
+      "step": 278
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9860808082638245e-05,
+      "loss": 1.8658,
+      "step": 279
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9858817233904872e-05,
+      "loss": 1.6387,
+      "step": 280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.985681234983623e-05,
+      "loss": 1.7362,
+      "step": 281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9854793433286544e-05,
+      "loss": 1.8246,
+      "step": 282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9852760487129998e-05,
+      "loss": 1.7408,
+      "step": 283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9850713514260772e-05,
+      "loss": 1.8302,
+      "step": 284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9848652517593005e-05,
+      "loss": 1.7418,
+      "step": 285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9846577500060786e-05,
+      "loss": 1.802,
+      "step": 286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9844488464618192e-05,
+      "loss": 1.6638,
+      "step": 287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.984238541423923e-05,
+      "loss": 1.9796,
+      "step": 288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9840268351917883e-05,
+      "loss": 2.012,
+      "step": 289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.983813728066806e-05,
+      "loss": 1.7997,
+      "step": 290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9835992203523636e-05,
+      "loss": 1.7265,
+      "step": 291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.98338331235384e-05,
+      "loss": 1.9192,
+      "step": 292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9831660043786097e-05,
+      "loss": 1.72,
+      "step": 293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.982947296736039e-05,
+      "loss": 1.984,
+      "step": 294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9827271897374883e-05,
+      "loss": 1.6382,
+      "step": 295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9825056836963078e-05,
+      "loss": 1.8195,
+      "step": 296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9822827789278412e-05,
+      "loss": 1.9058,
+      "step": 297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9820584757494234e-05,
+      "loss": 1.8352,
+      "step": 298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9818327744803796e-05,
+      "loss": 1.6767,
+      "step": 299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9816056754420253e-05,
+      "loss": 1.7036,
+      "step": 300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9813771789576663e-05,
+      "loss": 1.9956,
+      "step": 301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9811472853525978e-05,
+      "loss": 2.1392,
+      "step": 302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980915994954103e-05,
+      "loss": 1.9096,
+      "step": 303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980683308091456e-05,
+      "loss": 1.7885,
+      "step": 304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9804492250959164e-05,
+      "loss": 1.8036,
+      "step": 305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9802137463007327e-05,
+      "loss": 1.7883,
+      "step": 306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.97997687204114e-05,
+      "loss": 1.9368,
+      "step": 307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9797386026543607e-05,
+      "loss": 1.8897,
+      "step": 308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9794989384796025e-05,
+      "loss": 1.8347,
+      "step": 309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9792578798580597e-05,
+      "loss": 1.9128,
+      "step": 310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9790154271329105e-05,
+      "loss": 1.9121,
+      "step": 311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978771580649319e-05,
+      "loss": 1.9557,
+      "step": 312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978526340754433e-05,
+      "loss": 1.9909,
+      "step": 313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978279707797384e-05,
+      "loss": 1.7941,
+      "step": 314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978031682129287e-05,
+      "loss": 1.8897,
+      "step": 315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.977782264103239e-05,
+      "loss": 1.9731,
+      "step": 316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.97753145407432e-05,
+      "loss": 2.0019,
+      "step": 317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9772792523995912e-05,
+      "loss": 2.01,
+      "step": 318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9770256594380955e-05,
+      "loss": 1.9104,
+      "step": 319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.976770675550856e-05,
+      "loss": 1.9926,
+      "step": 320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9765143011008758e-05,
+      "loss": 2.1075,
+      "step": 321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9762565364531383e-05,
+      "loss": 1.916,
+      "step": 322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9759973819746055e-05,
+      "loss": 1.7118,
+      "step": 323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9757368380342185e-05,
+      "loss": 2.1362,
+      "step": 324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.975474905002896e-05,
+      "loss": 1.8019,
+      "step": 325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9752115832535343e-05,
+      "loss": 1.8218,
+      "step": 326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9749468731610072e-05,
+      "loss": 1.7612,
+      "step": 327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9746807751021645e-05,
+      "loss": 1.9373,
+      "step": 328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.974413289455832e-05,
+      "loss": 1.9847,
+      "step": 329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.974144416602811e-05,
+      "loss": 1.7258,
+      "step": 330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9738741569258782e-05,
+      "loss": 1.7487,
+      "step": 331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9736025108097834e-05,
+      "loss": 1.9034,
+      "step": 332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9733294786412513e-05,
+      "loss": 1.7912,
+      "step": 333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9730550608089795e-05,
+      "loss": 1.8738,
+      "step": 334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9727792577036383e-05,
+      "loss": 1.8648,
+      "step": 335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9725020697178697e-05,
+      "loss": 1.9759,
+      "step": 336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.972223497246288e-05,
+      "loss": 1.9795,
+      "step": 337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9719435406854775e-05,
+      "loss": 1.9294,
+      "step": 338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9716622004339943e-05,
+      "loss": 2.0429,
+      "step": 339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.971379476892363e-05,
+      "loss": 2.1394,
+      "step": 340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9710953704630784e-05,
+      "loss": 1.9411,
+      "step": 341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9708098815506035e-05,
+      "loss": 1.8354,
+      "step": 342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.97052301056137e-05,
+      "loss": 1.9368,
+      "step": 343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9702347579037765e-05,
+      "loss": 1.6728,
+      "step": 344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.969945123988189e-05,
+      "loss": 1.789,
+      "step": 345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9696541092269402e-05,
+      "loss": 1.7972,
+      "step": 346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9693617140343277e-05,
+      "loss": 2.2036,
+      "step": 347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.969067938826615e-05,
+      "loss": 1.796,
+      "step": 348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9687727840220304e-05,
+      "loss": 1.8355,
+      "step": 349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9684762500407662e-05,
+      "loss": 1.8797,
+      "step": 350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9681783373049768e-05,
+      "loss": 1.8698,
+      "step": 351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9678790462387816e-05,
+      "loss": 1.9311,
+      "step": 352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.967578377268261e-05,
+      "loss": 1.6739,
+      "step": 353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9672763308214566e-05,
+      "loss": 1.9781,
+      "step": 354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9669729073283724e-05,
+      "loss": 1.9173,
+      "step": 355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9666681072209712e-05,
+      "loss": 1.7903,
+      "step": 356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.966361930933177e-05,
+      "loss": 1.6572,
+      "step": 357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9660543789008724e-05,
+      "loss": 1.893,
+      "step": 358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9657454515618984e-05,
+      "loss": 1.6262,
+      "step": 359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.965435149356054e-05,
+      "loss": 1.8287,
+      "step": 360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9651234727250958e-05,
+      "loss": 1.924,
+      "step": 361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9648104221127363e-05,
+      "loss": 1.9815,
+      "step": 362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9644959979646455e-05,
+      "loss": 1.9731,
+      "step": 363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.964180200728447e-05,
+      "loss": 1.613,
+      "step": 364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9638630308537213e-05,
+      "loss": 1.6979,
+      "step": 365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9635444887920006e-05,
+      "loss": 1.7188,
+      "step": 366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9632245749967723e-05,
+      "loss": 1.9756,
+      "step": 367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9629032899234763e-05,
+      "loss": 2.0186,
+      "step": 368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9625806340295047e-05,
+      "loss": 1.952,
+      "step": 369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9622566077742e-05,
+      "loss": 1.6013,
+      "step": 370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.961931211618858e-05,
+      "loss": 1.8473,
+      "step": 371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9616044460267224e-05,
+      "loss": 1.9991,
+      "step": 372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9612763114629875e-05,
+      "loss": 1.6916,
+      "step": 373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.960946808394797e-05,
+      "loss": 2.0116,
+      "step": 374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9606159372912415e-05,
+      "loss": 2.0294,
+      "step": 375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9602836986233603e-05,
+      "loss": 1.8217,
+      "step": 376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.959950092864139e-05,
+      "loss": 1.8566,
+      "step": 377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9596151204885103e-05,
+      "loss": 1.8843,
+      "step": 378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9592787819733513e-05,
+      "loss": 1.9598,
+      "step": 379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9589410777974852e-05,
+      "loss": 1.9737,
+      "step": 380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9586020084416777e-05,
+      "loss": 1.9526,
+      "step": 381
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9582615743886397e-05,
+      "loss": 1.9623,
+      "step": 382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9579197761230243e-05,
+      "loss": 1.9201,
+      "step": 383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9575766141314264e-05,
+      "loss": 1.8925,
+      "step": 384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.957232088902383e-05,
+      "loss": 2.0269,
+      "step": 385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9568862009263712e-05,
+      "loss": 1.9761,
+      "step": 386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9565389506958086e-05,
+      "loss": 1.7009,
+      "step": 387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.956190338705052e-05,
+      "loss": 1.8927,
+      "step": 388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.955840365450397e-05,
+      "loss": 1.7908,
+      "step": 389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9554890314300768e-05,
+      "loss": 1.6918,
+      "step": 390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.955136337144262e-05,
+      "loss": 2.0751,
+      "step": 391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9547822830950597e-05,
+      "loss": 1.9847,
+      "step": 392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.954426869786513e-05,
+      "loss": 1.8185,
+      "step": 393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9540700977246e-05,
+      "loss": 1.8964,
+      "step": 394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9537119674172332e-05,
+      "loss": 1.7804,
+      "step": 395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.953352479374258e-05,
+      "loss": 2.0158,
+      "step": 396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9529916341074538e-05,
+      "loss": 1.7388,
+      "step": 397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9526294321305323e-05,
+      "loss": 1.8575,
+      "step": 398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9522658739591348e-05,
+      "loss": 1.8096,
+      "step": 399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9519009601108358e-05,
+      "loss": 1.7633,
+      "step": 400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.951534691105138e-05,
+      "loss": 1.6767,
+      "step": 401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9511670674634745e-05,
+      "loss": 1.7279,
+      "step": 402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9507980897092055e-05,
+      "loss": 1.7844,
+      "step": 403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9504277583676204e-05,
+      "loss": 1.8847,
+      "step": 404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.950056073965935e-05,
+      "loss": 1.7377,
+      "step": 405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9496830370332916e-05,
+      "loss": 1.7804,
+      "step": 406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.949308648100757e-05,
+      "loss": 1.6626,
+      "step": 407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9489329077013244e-05,
+      "loss": 1.8978,
+      "step": 408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.94855581636991e-05,
+      "loss": 2.005,
+      "step": 409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.948177374643353e-05,
+      "loss": 1.8455,
+      "step": 410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9477975830604158e-05,
+      "loss": 1.7812,
+      "step": 411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.947416442161782e-05,
+      "loss": 1.7934,
+      "step": 412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.947033952490056e-05,
+      "loss": 2.0325,
+      "step": 413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.946650114589763e-05,
+      "loss": 1.7311,
+      "step": 414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9462649290073473e-05,
+      "loss": 1.7788,
+      "step": 415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9458783962911717e-05,
+      "loss": 2.0093,
+      "step": 416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9454905169915164e-05,
+      "loss": 1.8025,
+      "step": 417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.945101291660579e-05,
+      "loss": 1.7943,
+      "step": 418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9447107208524744e-05,
+      "loss": 2.1016,
+      "step": 419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.944318805123231e-05,
+      "loss": 1.917,
+      "step": 420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.943925545030794e-05,
+      "loss": 1.8536,
+      "step": 421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9435309411350195e-05,
+      "loss": 1.9451,
+      "step": 422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9431349939976804e-05,
+      "loss": 1.8264,
+      "step": 423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.942737704182459e-05,
+      "loss": 1.895,
+      "step": 424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9423390722549506e-05,
+      "loss": 2.0386,
+      "step": 425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.94193909878266e-05,
+      "loss": 1.8459,
+      "step": 426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9415377843350032e-05,
+      "loss": 1.6088,
+      "step": 427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9411351294833047e-05,
+      "loss": 1.9419,
+      "step": 428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9407311348007965e-05,
+      "loss": 1.5992,
+      "step": 429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.940325800862619e-05,
+      "loss": 1.8278,
+      "step": 430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.939919128245819e-05,
+      "loss": 1.8029,
+      "step": 431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9395111175293485e-05,
+      "loss": 1.861,
+      "step": 432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.939101769294066e-05,
+      "loss": 1.8148,
+      "step": 433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9386910841227315e-05,
+      "loss": 1.7924,
+      "step": 434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9382790626000112e-05,
+      "loss": 1.829,
+      "step": 435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9378657053124718e-05,
+      "loss": 1.685,
+      "step": 436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9374510128485824e-05,
+      "loss": 2.1904,
+      "step": 437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9370349857987127e-05,
+      "loss": 1.8356,
+      "step": 438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9366176247551327e-05,
+      "loss": 1.9149,
+      "step": 439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.936198930312011e-05,
+      "loss": 1.6392,
+      "step": 440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9357789030654146e-05,
+      "loss": 1.7937,
+      "step": 441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9353575436133086e-05,
+      "loss": 1.8761,
+      "step": 442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9349348525555537e-05,
+      "loss": 1.9128,
+      "step": 443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9345108304939065e-05,
+      "loss": 1.9687,
+      "step": 444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9340854780320196e-05,
+      "loss": 1.8992,
+      "step": 445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9336587957754382e-05,
+      "loss": 1.8379,
+      "step": 446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9332307843316004e-05,
+      "loss": 1.7322,
+      "step": 447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9328014443098385e-05,
+      "loss": 1.7952,
+      "step": 448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.932370776321374e-05,
+      "loss": 1.7826,
+      "step": 449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.931938780979321e-05,
+      "loss": 1.9815,
+      "step": 450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9315054588986815e-05,
+      "loss": 1.5621,
+      "step": 451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9310708106963473e-05,
+      "loss": 1.7337,
+      "step": 452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.930634836991098e-05,
+      "loss": 1.8992,
+      "step": 453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9301975384035994e-05,
+      "loss": 1.7118,
+      "step": 454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.929758915556404e-05,
+      "loss": 1.6617,
+      "step": 455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9293189690739512e-05,
+      "loss": 1.9588,
+      "step": 456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9288776995825616e-05,
+      "loss": 1.5833,
+      "step": 457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9284351077104414e-05,
+      "loss": 1.7649,
+      "step": 458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9279911940876793e-05,
+      "loss": 1.714,
+      "step": 459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.927545959346245e-05,
+      "loss": 1.9171,
+      "step": 460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.927099404119989e-05,
+      "loss": 1.6816,
+      "step": 461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9266515290446422e-05,
+      "loss": 1.6948,
+      "step": 462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.926202334757814e-05,
+      "loss": 1.9208,
+      "step": 463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9257518218989925e-05,
+      "loss": 2.0632,
+      "step": 464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9252999911095425e-05,
+      "loss": 1.9065,
+      "step": 465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.924846843032705e-05,
+      "loss": 1.8936,
+      "step": 466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9243923783135963e-05,
+      "loss": 1.9589,
+      "step": 467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9239365975992075e-05,
+      "loss": 2.0159,
+      "step": 468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.923479501538403e-05,
+      "loss": 1.8736,
+      "step": 469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9230210907819194e-05,
+      "loss": 1.8036,
+      "step": 470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9225613659823653e-05,
+      "loss": 1.8388,
+      "step": 471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9221003277942206e-05,
+      "loss": 2.069,
+      "step": 472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9216379768738338e-05,
+      "loss": 1.9428,
+      "step": 473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.921174313879423e-05,
+      "loss": 1.7716,
+      "step": 474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9207093394710742e-05,
+      "loss": 1.8555,
+      "step": 475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.920243054310739e-05,
+      "loss": 2.1768,
+      "step": 476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9197754590622385e-05,
+      "loss": 1.8459,
+      "step": 477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9193065543912546e-05,
+      "loss": 1.8115,
+      "step": 478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9188363409653363e-05,
+      "loss": 1.8124,
+      "step": 479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9183648194538946e-05,
+      "loss": 1.7224,
+      "step": 480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9178919905282033e-05,
+      "loss": 1.8167,
+      "step": 481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9174178548613967e-05,
+      "loss": 1.8557,
+      "step": 482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9169424131284698e-05,
+      "loss": 1.7496,
+      "step": 483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9164656660062773e-05,
+      "loss": 1.8104,
+      "step": 484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9159876141735324e-05,
+      "loss": 1.7362,
+      "step": 485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9155082583108048e-05,
+      "loss": 2.0344,
+      "step": 486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915027599100521e-05,
+      "loss": 1.7144,
+      "step": 487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9145456372269643e-05,
+      "loss": 1.8579,
+      "step": 488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.91406237337627e-05,
+      "loss": 1.7571,
+      "step": 489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9135778082364295e-05,
+      "loss": 1.5567,
+      "step": 490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.913091942497285e-05,
+      "loss": 1.7423,
+      "step": 491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9126047768505307e-05,
+      "loss": 1.9152,
+      "step": 492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9121163119897122e-05,
+      "loss": 1.7091,
+      "step": 493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9116265486102237e-05,
+      "loss": 1.7242,
+      "step": 494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9111354874093087e-05,
+      "loss": 1.9924,
+      "step": 495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.910643129086058e-05,
+      "loss": 1.8313,
+      "step": 496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9101494743414082e-05,
+      "loss": 1.8238,
+      "step": 497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9096545238781433e-05,
+      "loss": 1.987,
+      "step": 498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9091582784008908e-05,
+      "loss": 1.8133,
+      "step": 499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.908660738616122e-05,
+      "loss": 1.8589,
+      "step": 500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.908161905232151e-05,
+      "loss": 1.8214,
+      "step": 501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9076617789591326e-05,
+      "loss": 1.7285,
+      "step": 502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9071603605090638e-05,
+      "loss": 1.8021,
+      "step": 503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9066576505957796e-05,
+      "loss": 1.7359,
+      "step": 504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9061536499349553e-05,
+      "loss": 2.1337,
+      "step": 505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.905648359244102e-05,
+      "loss": 1.787,
+      "step": 506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.905141779242568e-05,
+      "loss": 1.7364,
+      "step": 507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.904633910651538e-05,
+      "loss": 1.6271,
+      "step": 508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9041247541940294e-05,
+      "loss": 1.9234,
+      "step": 509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9036143105948944e-05,
+      "loss": 2.0125,
+      "step": 510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9031025805808175e-05,
+      "loss": 2.0037,
+      "step": 511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.902589564880314e-05,
+      "loss": 1.4813,
+      "step": 512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9020752642237302e-05,
+      "loss": 1.8722,
+      "step": 513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9015596793432408e-05,
+      "loss": 1.9763,
+      "step": 514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9010428109728497e-05,
+      "loss": 1.7076,
+      "step": 515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9005246598483876e-05,
+      "loss": 1.7394,
+      "step": 516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9000052267075116e-05,
+      "loss": 1.6068,
+      "step": 517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8994845122897033e-05,
+      "loss": 1.8676,
+      "step": 518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.898962517336269e-05,
+      "loss": 1.6328,
+      "step": 519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8984392425903384e-05,
+      "loss": 2.1272,
+      "step": 520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8979146887968615e-05,
+      "loss": 1.7898,
+      "step": 521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.897388856702611e-05,
+      "loss": 1.8246,
+      "step": 522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8968617470561788e-05,
+      "loss": 1.6391,
+      "step": 523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.896333360607975e-05,
+      "loss": 1.6649,
+      "step": 524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895803698110228e-05,
+      "loss": 1.7837,
+      "step": 525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895272760316983e-05,
+      "loss": 1.7271,
+      "step": 526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8947405479840998e-05,
+      "loss": 1.6772,
+      "step": 527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894207061869254e-05,
+      "loss": 2.2271,
+      "step": 528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8936723027319334e-05,
+      "loss": 1.6941,
+      "step": 529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8931362713334393e-05,
+      "loss": 1.844,
+      "step": 530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8925989684368823e-05,
+      "loss": 1.5838,
+      "step": 531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.892060394807186e-05,
+      "loss": 2.196,
+      "step": 532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8915205512110802e-05,
+      "loss": 1.8994,
+      "step": 533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8909794384171048e-05,
+      "loss": 2.097,
+      "step": 534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8904370571956053e-05,
+      "loss": 1.8567,
+      "step": 535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.889893408318733e-05,
+      "loss": 1.7573,
+      "step": 536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.889348492560445e-05,
+      "loss": 1.8545,
+      "step": 537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8888023106965007e-05,
+      "loss": 1.8982,
+      "step": 538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8882548635044625e-05,
+      "loss": 1.8322,
+      "step": 539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8877061517636948e-05,
+      "loss": 1.7116,
+      "step": 540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8871561762553612e-05,
+      "loss": 1.8873,
+      "step": 541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8866049377624245e-05,
+      "loss": 1.785,
+      "step": 542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8860524370696464e-05,
+      "loss": 1.9869,
+      "step": 543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8854986749635847e-05,
+      "loss": 2.0405,
+      "step": 544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.884943652232593e-05,
+      "loss": 1.7545,
+      "step": 545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8843873696668208e-05,
+      "loss": 1.9384,
+      "step": 546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8838298280582097e-05,
+      "loss": 1.7382,
+      "step": 547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8832710282004936e-05,
+      "loss": 1.9079,
+      "step": 548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.882710970889199e-05,
+      "loss": 1.7966,
+      "step": 549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.882149656921642e-05,
+      "loss": 1.7548,
+      "step": 550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8815870870969267e-05,
+      "loss": 2.0582,
+      "step": 551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8810232622159465e-05,
+      "loss": 1.9251,
+      "step": 552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8804581830813812e-05,
+      "loss": 1.7577,
+      "step": 553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8798918504976953e-05,
+      "loss": 1.8336,
+      "step": 554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8793242652711388e-05,
+      "loss": 2.0418,
+      "step": 555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8787554282097452e-05,
+      "loss": 1.7,
+      "step": 556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8781853401233285e-05,
+      "loss": 1.9158,
+      "step": 557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8776140018234855e-05,
+      "loss": 1.5324,
+      "step": 558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8770414141235918e-05,
+      "loss": 1.934,
+      "step": 559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8764675778388023e-05,
+      "loss": 1.9987,
+      "step": 560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8758924937860487e-05,
+      "loss": 1.7242,
+      "step": 561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.87531616278404e-05,
+      "loss": 1.7915,
+      "step": 562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8747385856532596e-05,
+      "loss": 2.0531,
+      "step": 563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.874159763215965e-05,
+      "loss": 2.0786,
+      "step": 564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8735796962961878e-05,
+      "loss": 1.8106,
+      "step": 565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.872998385719729e-05,
+      "loss": 1.8355,
+      "step": 566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.872415832314162e-05,
+      "loss": 1.7494,
+      "step": 567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.871832036908829e-05,
+      "loss": 1.9507,
+      "step": 568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.87124700033484e-05,
+      "loss": 1.8078,
+      "step": 569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8706607234250723e-05,
+      "loss": 1.8325,
+      "step": 570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8700732070141693e-05,
+      "loss": 1.8502,
+      "step": 571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8694844519385383e-05,
+      "loss": 1.7116,
+      "step": 572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.86889445903635e-05,
+      "loss": 1.8984,
+      "step": 573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8683032291475382e-05,
+      "loss": 1.8425,
+      "step": 574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.867710763113797e-05,
+      "loss": 1.7594,
+      "step": 575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.86711706177858e-05,
+      "loss": 1.6792,
+      "step": 576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8665221259871005e-05,
+      "loss": 1.9078,
+      "step": 577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.865925956586328e-05,
+      "loss": 1.6609,
+      "step": 578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8653285544249896e-05,
+      "loss": 1.7867,
+      "step": 579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.864729920353566e-05,
+      "loss": 1.7511,
+      "step": 580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.864130055224292e-05,
+      "loss": 1.9401,
+      "step": 581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.863528959891156e-05,
+      "loss": 2.0443,
+      "step": 582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8629266352098964e-05,
+      "loss": 1.8604,
+      "step": 583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8623230820380026e-05,
+      "loss": 1.6141,
+      "step": 584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.861718301234713e-05,
+      "loss": 2.0134,
+      "step": 585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.861112293661013e-05,
+      "loss": 1.5974,
+      "step": 586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8605050601796345e-05,
+      "loss": 1.675,
+      "step": 587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8598966016550556e-05,
+      "loss": 1.6918,
+      "step": 588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8592869189534974e-05,
+      "loss": 1.7935,
+      "step": 589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8586760129429247e-05,
+      "loss": 2.0052,
+      "step": 590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8580638844930425e-05,
+      "loss": 1.8375,
+      "step": 591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8574505344752977e-05,
+      "loss": 2.0357,
+      "step": 592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.856835963762875e-05,
+      "loss": 1.9813,
+      "step": 593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8562201732306976e-05,
+      "loss": 1.8431,
+      "step": 594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8556031637554254e-05,
+      "loss": 1.922,
+      "step": 595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8549849362154524e-05,
+      "loss": 1.7421,
+      "step": 596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8543654914909083e-05,
+      "loss": 1.8718,
+      "step": 597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8537448304636546e-05,
+      "loss": 1.943,
+      "step": 598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8531229540172852e-05,
+      "loss": 1.9991,
+      "step": 599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.852499863037123e-05,
+      "loss": 1.6649,
+      "step": 600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8518755584102214e-05,
+      "loss": 1.8183,
+      "step": 601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8512500410253604e-05,
+      "loss": 1.7537,
+      "step": 602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8506233117730478e-05,
+      "loss": 1.8279,
+      "step": 603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8499953715455148e-05,
+      "loss": 1.6765,
+      "step": 604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8493662212367184e-05,
+      "loss": 1.9644,
+      "step": 605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8487358617423376e-05,
+      "loss": 2.0155,
+      "step": 606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8481042939597727e-05,
+      "loss": 1.7584,
+      "step": 607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8474715187881444e-05,
+      "loss": 1.6792,
+      "step": 608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.846837537128292e-05,
+      "loss": 1.8875,
+      "step": 609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8462023498827728e-05,
+      "loss": 1.7411,
+      "step": 610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.84556595795586e-05,
+      "loss": 1.4576,
+      "step": 611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8449283622535416e-05,
+      "loss": 1.7889,
+      "step": 612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8442895636835205e-05,
+      "loss": 1.7446,
+      "step": 613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.843649563155211e-05,
+      "loss": 2.0262,
+      "step": 614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8430083615797386e-05,
+      "loss": 1.7133,
+      "step": 615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.842365959869939e-05,
+      "loss": 2.1312,
+      "step": 616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8417223589403567e-05,
+      "loss": 2.0981,
+      "step": 617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8410775597072418e-05,
+      "loss": 1.6955,
+      "step": 618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8404315630885535e-05,
+      "loss": 1.5846,
+      "step": 619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8397843700039523e-05,
+      "loss": 1.6552,
+      "step": 620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.839135981374804e-05,
+      "loss": 1.7879,
+      "step": 621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.838486398124176e-05,
+      "loss": 1.7106,
+      "step": 622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8378356211768364e-05,
+      "loss": 1.8767,
+      "step": 623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.837183651459252e-05,
+      "loss": 2.015,
+      "step": 624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8365304898995887e-05,
+      "loss": 1.5223,
+      "step": 625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8358761374277088e-05,
+      "loss": 2.0143,
+      "step": 626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8352205949751695e-05,
+      "loss": 1.725,
+      "step": 627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8345638634752227e-05,
+      "loss": 1.841,
+      "step": 628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8339059438628134e-05,
+      "loss": 1.7485,
+      "step": 629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8332468370745766e-05,
+      "loss": 1.5277,
+      "step": 630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.832586544048839e-05,
+      "loss": 1.6906,
+      "step": 631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.831925065725615e-05,
+      "loss": 1.7929,
+      "step": 632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.831262403046607e-05,
+      "loss": 1.6509,
+      "step": 633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8305985569552034e-05,
+      "loss": 2.0475,
+      "step": 634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.829933528396477e-05,
+      "loss": 1.9633,
+      "step": 635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8292673183171845e-05,
+      "loss": 1.8058,
+      "step": 636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8285999276657642e-05,
+      "loss": 1.7511,
+      "step": 637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8279313573923354e-05,
+      "loss": 1.9296,
+      "step": 638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8272616084486968e-05,
+      "loss": 1.8956,
+      "step": 639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8265906817883244e-05,
+      "loss": 1.8134,
+      "step": 640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.825918578366372e-05,
+      "loss": 2.0214,
+      "step": 641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8252452991396676e-05,
+      "loss": 1.7222,
+      "step": 642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.824570845066714e-05,
+      "loss": 1.8207,
+      "step": 643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8238952171076862e-05,
+      "loss": 1.6518,
+      "step": 644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8232184162244297e-05,
+      "loss": 1.5583,
+      "step": 645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8225404433804605e-05,
+      "loss": 2.127,
+      "step": 646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.821861299540963e-05,
+      "loss": 1.561,
+      "step": 647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8211809856727892e-05,
+      "loss": 1.9207,
+      "step": 648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8204995027444554e-05,
+      "loss": 1.9223,
+      "step": 649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.819816851726144e-05,
+      "loss": 1.8233,
+      "step": 650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8191330335896985e-05,
+      "loss": 1.4765,
+      "step": 651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8184480493086248e-05,
+      "loss": 1.6904,
+      "step": 652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8177618998580897e-05,
+      "loss": 1.9904,
+      "step": 653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8170745862149174e-05,
+      "loss": 1.8617,
+      "step": 654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8163861093575905e-05,
+      "loss": 2.1474,
+      "step": 655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.815696470266247e-05,
+      "loss": 1.967,
+      "step": 656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8150056699226793e-05,
+      "loss": 1.7864,
+      "step": 657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8143137093103345e-05,
+      "loss": 1.8498,
+      "step": 658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8136205894143098e-05,
+      "loss": 1.8173,
+      "step": 659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8129263112213527e-05,
+      "loss": 1.9093,
+      "step": 660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8122308757198614e-05,
+      "loss": 1.4915,
+      "step": 661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8115342838998807e-05,
+      "loss": 1.8303,
+      "step": 662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8108365367531003e-05,
+      "loss": 1.7912,
+      "step": 663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8101376352728572e-05,
+      "loss": 1.7086,
+      "step": 664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.80943758045413e-05,
+      "loss": 2.0016,
+      "step": 665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8087363732935398e-05,
+      "loss": 1.7456,
+      "step": 666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8080340147893477e-05,
+      "loss": 1.7095,
+      "step": 667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.807330505941455e-05,
+      "loss": 1.8166,
+      "step": 668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8066258477513992e-05,
+      "loss": 1.9571,
+      "step": 669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.805920041222355e-05,
+      "loss": 1.857,
+      "step": 670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.805213087359132e-05,
+      "loss": 1.8744,
+      "step": 671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8045049871681726e-05,
+      "loss": 1.8435,
+      "step": 672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8037957416575514e-05,
+      "loss": 2.0229,
+      "step": 673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.803085351836974e-05,
+      "loss": 1.8561,
+      "step": 674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.802373818717774e-05,
+      "loss": 1.5955,
+      "step": 675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8016611433129135e-05,
+      "loss": 1.6131,
+      "step": 676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8009473266369806e-05,
+      "loss": 1.8665,
+      "step": 677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8002323697061883e-05,
+      "loss": 1.9577,
+      "step": 678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7995162735383725e-05,
+      "loss": 1.6646,
+      "step": 679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.798799039152991e-05,
+      "loss": 1.7852,
+      "step": 680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7980806675711225e-05,
+      "loss": 1.904,
+      "step": 681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7973611598154644e-05,
+      "loss": 1.7585,
+      "step": 682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7966405169103313e-05,
+      "loss": 1.8896,
+      "step": 683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.795918739881654e-05,
+      "loss": 1.7932,
+      "step": 684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7951958297569775e-05,
+      "loss": 1.887,
+      "step": 685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7944717875654615e-05,
+      "loss": 1.7784,
+      "step": 686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7937466143378754e-05,
+      "loss": 1.598,
+      "step": 687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7930203111065997e-05,
+      "loss": 1.8335,
+      "step": 688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7922928789056233e-05,
+      "loss": 1.7218,
+      "step": 689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7915643187705428e-05,
+      "loss": 1.8661,
+      "step": 690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7908346317385602e-05,
+      "loss": 2.0096,
+      "step": 691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7901038188484818e-05,
+      "loss": 1.9253,
+      "step": 692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.789371881140717e-05,
+      "loss": 1.9858,
+      "step": 693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7886388196572758e-05,
+      "loss": 1.6172,
+      "step": 694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.787904635441769e-05,
+      "loss": 2.1303,
+      "step": 695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.787169329539405e-05,
+      "loss": 1.9885,
+      "step": 696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7864329029969903e-05,
+      "loss": 1.8695,
+      "step": 697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7856953568629243e-05,
+      "loss": 1.816,
+      "step": 698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.784956692187203e-05,
+      "loss": 1.7575,
+      "step": 699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7842169100214137e-05,
+      "loss": 1.9882,
+      "step": 700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7834760114187334e-05,
+      "loss": 1.8572,
+      "step": 701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.782733997433931e-05,
+      "loss": 2.1264,
+      "step": 702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.781990869123361e-05,
+      "loss": 1.7489,
+      "step": 703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7812466275449656e-05,
+      "loss": 1.9035,
+      "step": 704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7805012737582713e-05,
+      "loss": 2.0032,
+      "step": 705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.779754808824388e-05,
+      "loss": 1.7685,
+      "step": 706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7790072338060086e-05,
+      "loss": 2.076,
+      "step": 707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7782585497674043e-05,
+      "loss": 2.145,
+      "step": 708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7775087577744266e-05,
+      "loss": 1.8889,
+      "step": 709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.776757858894504e-05,
+      "loss": 1.7354,
+      "step": 710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7760058541966406e-05,
+      "loss": 1.8177,
+      "step": 711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7752527447514154e-05,
+      "loss": 1.8914,
+      "step": 712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7744985316309793e-05,
+      "loss": 1.5651,
+      "step": 713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.773743215909055e-05,
+      "loss": 1.8058,
+      "step": 714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7729867986609338e-05,
+      "loss": 2.0572,
+      "step": 715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.772229280963478e-05,
+      "loss": 1.8813,
+      "step": 716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7714706638951126e-05,
+      "loss": 1.8782,
+      "step": 717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.770710948535831e-05,
+      "loss": 1.9337,
+      "step": 718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.769950135967188e-05,
+      "loss": 1.9506,
+      "step": 719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7691882272723023e-05,
+      "loss": 1.8569,
+      "step": 720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7684252235358514e-05,
+      "loss": 2.0687,
+      "step": 721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7676611258440726e-05,
+      "loss": 2.0012,
+      "step": 722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.76689593528476e-05,
+      "loss": 1.6913,
+      "step": 723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.766129652947265e-05,
+      "loss": 1.7295,
+      "step": 724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7653622799224914e-05,
+      "loss": 1.4796,
+      "step": 725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7645938173028964e-05,
+      "loss": 1.8595,
+      "step": 726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7638242661824892e-05,
+      "loss": 1.7979,
+      "step": 727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7630536276568277e-05,
+      "loss": 2.0496,
+      "step": 728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.762281902823018e-05,
+      "loss": 1.7305,
+      "step": 729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7615090927797135e-05,
+      "loss": 1.7799,
+      "step": 730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.760735198627111e-05,
+      "loss": 1.8931,
+      "step": 731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7599602214669522e-05,
+      "loss": 1.9162,
+      "step": 732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.75918416240252e-05,
+      "loss": 1.8491,
+      "step": 733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7584070225386368e-05,
+      "loss": 1.9094,
+      "step": 734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7576288029816654e-05,
+      "loss": 1.6776,
+      "step": 735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.756849504839504e-05,
+      "loss": 1.7776,
+      "step": 736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7560691292215872e-05,
+      "loss": 1.9614,
+      "step": 737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7552876772388833e-05,
+      "loss": 1.9188,
+      "step": 738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7545051500038926e-05,
+      "loss": 1.6998,
+      "step": 739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.753721548630647e-05,
+      "loss": 1.7554,
+      "step": 740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7529368742347066e-05,
+      "loss": 1.7072,
+      "step": 741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.75215112793316e-05,
+      "loss": 1.6595,
+      "step": 742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7513643108446213e-05,
+      "loss": 1.6982,
+      "step": 743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.750576424089229e-05,
+      "loss": 1.5064,
+      "step": 744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7497874687886447e-05,
+      "loss": 1.7888,
+      "step": 745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7489974460660507e-05,
+      "loss": 1.6298,
+      "step": 746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7482063570461493e-05,
+      "loss": 1.9298,
+      "step": 747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.747414202855161e-05,
+      "loss": 1.8131,
+      "step": 748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7466209846208225e-05,
+      "loss": 1.7347,
+      "step": 749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7458267034723846e-05,
+      "loss": 1.7985,
+      "step": 750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745031360540613e-05,
+      "loss": 1.6756,
+      "step": 751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744234956957783e-05,
+      "loss": 1.5744,
+      "step": 752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743437493857681e-05,
+      "loss": 2.2723,
+      "step": 753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7426389723756026e-05,
+      "loss": 1.8948,
+      "step": 754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741839393648348e-05,
+      "loss": 2.0171,
+      "step": 755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741038758814224e-05,
+      "loss": 2.0134,
+      "step": 756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7402370690130406e-05,
+      "loss": 2.0325,
+      "step": 757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.73943432538611e-05,
+      "loss": 1.7718,
+      "step": 758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7386305290762437e-05,
+      "loss": 1.9268,
+      "step": 759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737825681227753e-05,
+      "loss": 2.1387,
+      "step": 760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7370197829864454e-05,
+      "loss": 1.7365,
+      "step": 761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7362128354996242e-05,
+      "loss": 1.876,
+      "step": 762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7354048399160866e-05,
+      "loss": 1.9904,
+      "step": 763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7345957973861205e-05,
+      "loss": 2.1336,
+      "step": 764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7337857090615068e-05,
+      "loss": 1.5848,
+      "step": 765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7329745760955122e-05,
+      "loss": 1.6669,
+      "step": 766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.732162399642894e-05,
+      "loss": 1.73,
+      "step": 767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7313491808598914e-05,
+      "loss": 1.9249,
+      "step": 768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7305349209042303e-05,
+      "loss": 1.6414,
+      "step": 769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.729719620935118e-05,
+      "loss": 1.7314,
+      "step": 770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.728903282113242e-05,
+      "loss": 1.7717,
+      "step": 771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7280859056007682e-05,
+      "loss": 1.8558,
+      "step": 772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7272674925613424e-05,
+      "loss": 2.0347,
+      "step": 773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7264480441600823e-05,
+      "loss": 1.7946,
+      "step": 774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7256275615635826e-05,
+      "loss": 1.5202,
+      "step": 775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7248060459399092e-05,
+      "loss": 1.5847,
+      "step": 776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7239834984585982e-05,
+      "loss": 1.7596,
+      "step": 777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7231599202906553e-05,
+      "loss": 1.5023,
+      "step": 778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7223353126085536e-05,
+      "loss": 1.6659,
+      "step": 779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7215096765862315e-05,
+      "loss": 1.8749,
+      "step": 780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.720683013399091e-05,
+      "loss": 1.806,
+      "step": 781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7198553242239978e-05,
+      "loss": 1.7427,
+      "step": 782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7190266102392765e-05,
+      "loss": 1.7963,
+      "step": 783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.718196872624712e-05,
+      "loss": 1.8034,
+      "step": 784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.717366112561546e-05,
+      "loss": 1.6718,
+      "step": 785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7165343312324755e-05,
+      "loss": 1.7927,
+      "step": 786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7157015298216516e-05,
+      "loss": 1.7335,
+      "step": 787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.714867709514678e-05,
+      "loss": 1.7174,
+      "step": 788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.714032871498608e-05,
+      "loss": 1.9039,
+      "step": 789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.713197016961945e-05,
+      "loss": 1.874,
+      "step": 790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7123601470946388e-05,
+      "loss": 1.7786,
+      "step": 791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7115222630880844e-05,
+      "loss": 1.8011,
+      "step": 792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7106833661351213e-05,
+      "loss": 2.0506,
+      "step": 793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7098434574300307e-05,
+      "loss": 2.0258,
+      "step": 794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7090025381685337e-05,
+      "loss": 1.628,
+      "step": 795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.708160609547791e-05,
+      "loss": 1.7226,
+      "step": 796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7073176727663994e-05,
+      "loss": 2.1619,
+      "step": 797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.706473729024392e-05,
+      "loss": 1.6363,
+      "step": 798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7056287795232338e-05,
+      "loss": 1.4952,
+      "step": 799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7047828254658233e-05,
+      "loss": 1.7071,
+      "step": 800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.703935868056488e-05,
+      "loss": 1.9266,
+      "step": 801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.703087908500985e-05,
+      "loss": 1.6527,
+      "step": 802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.702238948006496e-05,
+      "loss": 1.8177,
+      "step": 803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.70138898778163e-05,
+      "loss": 1.8119,
+      "step": 804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7005380290364182e-05,
+      "loss": 1.887,
+      "step": 805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6996860729823127e-05,
+      "loss": 1.5231,
+      "step": 806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6988331208321868e-05,
+      "loss": 1.8765,
+      "step": 807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6979791738003305e-05,
+      "loss": 2.113,
+      "step": 808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.697124233102451e-05,
+      "loss": 1.7483,
+      "step": 809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.69626829995567e-05,
+      "loss": 1.7028,
+      "step": 810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.695411375578522e-05,
+      "loss": 1.8967,
+      "step": 811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6945534611909525e-05,
+      "loss": 1.8347,
+      "step": 812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6936945580143166e-05,
+      "loss": 1.8635,
+      "step": 813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6928346672713768e-05,
+      "loss": 1.8776,
+      "step": 814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6919737901863024e-05,
+      "loss": 1.6324,
+      "step": 815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6911119279846655e-05,
+      "loss": 1.6854,
+      "step": 816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6902490818934417e-05,
+      "loss": 1.8578,
+      "step": 817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6893852531410066e-05,
+      "loss": 1.6869,
+      "step": 818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6885204429571356e-05,
+      "loss": 1.7316,
+      "step": 819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6876546525730005e-05,
+      "loss": 1.85,
+      "step": 820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.686787883221169e-05,
+      "loss": 1.6942,
+      "step": 821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6859201361356025e-05,
+      "loss": 1.7333,
+      "step": 822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.685051412551654e-05,
+      "loss": 1.4995,
+      "step": 823
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6841817137060666e-05,
+      "loss": 1.7027,
+      "step": 824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.683311040836973e-05,
+      "loss": 2.003,
+      "step": 825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.68243939518389e-05,
+      "loss": 1.7645,
+      "step": 826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6815667779877226e-05,
+      "loss": 1.6719,
+      "step": 827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6806931904907562e-05,
+      "loss": 1.8548,
+      "step": 828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6798186339366593e-05,
+      "loss": 1.7971,
+      "step": 829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.678943109570479e-05,
+      "loss": 1.8877,
+      "step": 830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.67806661863864e-05,
+      "loss": 1.7383,
+      "step": 831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.677189162388944e-05,
+      "loss": 1.7564,
+      "step": 832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.676310742070566e-05,
+      "loss": 1.6716,
+      "step": 833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6754313589340546e-05,
+      "loss": 1.6561,
+      "step": 834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.674551014231328e-05,
+      "loss": 2.1709,
+      "step": 835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.673669709215674e-05,
+      "loss": 1.5265,
+      "step": 836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6727874451417473e-05,
+      "loss": 1.8503,
+      "step": 837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6719042232655677e-05,
+      "loss": 1.8054,
+      "step": 838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.671020044844519e-05,
+      "loss": 1.8034,
+      "step": 839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6701349111373465e-05,
+      "loss": 1.5963,
+      "step": 840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6692488234041556e-05,
+      "loss": 1.9236,
+      "step": 841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.66836178290641e-05,
+      "loss": 1.5941,
+      "step": 842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.667473790906929e-05,
+      "loss": 1.8286,
+      "step": 843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.666584848669888e-05,
+      "loss": 1.7725,
+      "step": 844
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6656949574608138e-05,
+      "loss": 1.6444,
+      "step": 845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6648041185465846e-05,
+      "loss": 1.8188,
+      "step": 846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6639123331954276e-05,
+      "loss": 1.7729,
+      "step": 847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6630196026769187e-05,
+      "loss": 1.6913,
+      "step": 848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.662125928261977e-05,
+      "loss": 1.7507,
+      "step": 849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.661231311222868e-05,
+      "loss": 1.6705,
+      "step": 850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6603357528331966e-05,
+      "loss": 1.8839,
+      "step": 851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6594392543679098e-05,
+      "loss": 1.7405,
+      "step": 852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6585418171032925e-05,
+      "loss": 1.8388,
+      "step": 853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6576434423169647e-05,
+      "loss": 1.4925,
+      "step": 854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6567441312878828e-05,
+      "loss": 1.899,
+      "step": 855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6558438852963358e-05,
+      "loss": 2.1859,
+      "step": 856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6549427056239432e-05,
+      "loss": 1.7347,
+      "step": 857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6540405935536532e-05,
+      "loss": 1.8964,
+      "step": 858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6531375503697433e-05,
+      "loss": 1.9026,
+      "step": 859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6522335773578143e-05,
+      "loss": 1.7895,
+      "step": 860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6513286758047923e-05,
+      "loss": 2.135,
+      "step": 861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6504228469989248e-05,
+      "loss": 1.789,
+      "step": 862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6495160922297793e-05,
+      "loss": 2.0962,
+      "step": 863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6486084127882416e-05,
+      "loss": 1.8792,
+      "step": 864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.647699809966514e-05,
+      "loss": 1.6231,
+      "step": 865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.646790285058113e-05,
+      "loss": 1.5529,
+      "step": 866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6458798393578684e-05,
+      "loss": 1.8931,
+      "step": 867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.64496847416192e-05,
+      "loss": 1.5178,
+      "step": 868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.644056190767718e-05,
+      "loss": 1.791,
+      "step": 869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6431429904740183e-05,
+      "loss": 1.6673,
+      "step": 870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6422288745808828e-05,
+      "loss": 1.6321,
+      "step": 871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.641313844389677e-05,
+      "loss": 1.8554,
+      "step": 872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6403979012030677e-05,
+      "loss": 1.6654,
+      "step": 873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6394810463250218e-05,
+      "loss": 1.7586,
+      "step": 874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6385632810608035e-05,
+      "loss": 1.8075,
+      "step": 875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6376446067169744e-05,
+      "loss": 1.7,
+      "step": 876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6367250246013887e-05,
+      "loss": 1.8388,
+      "step": 877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6358045360231936e-05,
+      "loss": 1.562,
+      "step": 878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6348831422928277e-05,
+      "loss": 1.9606,
+      "step": 879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6339608447220163e-05,
+      "loss": 1.9655,
+      "step": 880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6330376446237726e-05,
+      "loss": 2.0021,
+      "step": 881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6321135433123946e-05,
+      "loss": 1.9258,
+      "step": 882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6311885421034638e-05,
+      "loss": 1.7109,
+      "step": 883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6302626423138412e-05,
+      "loss": 2.0604,
+      "step": 884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.629335845261669e-05,
+      "loss": 1.7187,
+      "step": 885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6284081522663652e-05,
+      "loss": 1.8782,
+      "step": 886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6274795646486244e-05,
+      "loss": 1.8045,
+      "step": 887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.626550083730414e-05,
+      "loss": 2.041,
+      "step": 888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6256197108349734e-05,
+      "loss": 1.8206,
+      "step": 889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6246884472868128e-05,
+      "loss": 1.7373,
+      "step": 890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6237562944117087e-05,
+      "loss": 1.6757,
+      "step": 891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6228232535367045e-05,
+      "loss": 1.9031,
+      "step": 892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.621889325990108e-05,
+      "loss": 1.626,
+      "step": 893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6209545131014893e-05,
+      "loss": 1.5336,
+      "step": 894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6200188162016778e-05,
+      "loss": 1.8612,
+      "step": 895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.619082236622763e-05,
+      "loss": 1.8652,
+      "step": 896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6181447756980903e-05,
+      "loss": 1.9462,
+      "step": 897
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6172064347622595e-05,
+      "loss": 1.7591,
+      "step": 898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.616267215151124e-05,
+      "loss": 2.0353,
+      "step": 899
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6153271182017864e-05,
+      "loss": 1.7371,
+      "step": 900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.614386145252601e-05,
+      "loss": 1.5204,
+      "step": 901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613444297643167e-05,
+      "loss": 1.6634,
+      "step": 902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6125015767143293e-05,
+      "loss": 1.9809,
+      "step": 903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.611557983808177e-05,
+      "loss": 1.9466,
+      "step": 904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.610613520268039e-05,
+      "loss": 1.8158,
+      "step": 905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.609668187438485e-05,
+      "loss": 1.6599,
+      "step": 906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.608721986665322e-05,
+      "loss": 1.6313,
+      "step": 907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6077749192955922e-05,
+      "loss": 2.0018,
+      "step": 908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6068269866775722e-05,
+      "loss": 1.5188,
+      "step": 909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6058781901607687e-05,
+      "loss": 1.901,
+      "step": 910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604928531095921e-05,
+      "loss": 1.7248,
+      "step": 911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6039780108349938e-05,
+      "loss": 1.9463,
+      "step": 912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603026630731179e-05,
+      "loss": 1.9392,
+      "step": 913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6020743921388937e-05,
+      "loss": 1.7979,
+      "step": 914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6011212964137742e-05,
+      "loss": 1.9517,
+      "step": 915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.60016734491268e-05,
+      "loss": 1.5702,
+      "step": 916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5992125389936873e-05,
+      "loss": 1.775,
+      "step": 917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.598256880016089e-05,
+      "loss": 1.8394,
+      "step": 918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5973003693403928e-05,
+      "loss": 1.7364,
+      "step": 919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5963430083283185e-05,
+      "loss": 2.0016,
+      "step": 920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5953847983427967e-05,
+      "loss": 1.7369,
+      "step": 921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5944257407479665e-05,
+      "loss": 2.1433,
+      "step": 922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5934658369091734e-05,
+      "loss": 1.8075,
+      "step": 923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5925050881929682e-05,
+      "loss": 1.7377,
+      "step": 924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5915434959671046e-05,
+      "loss": 1.6292,
+      "step": 925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.590581061600536e-05,
+      "loss": 1.5613,
+      "step": 926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5896177864634166e-05,
+      "loss": 1.7109,
+      "step": 927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5886536719270956e-05,
+      "loss": 1.4451,
+      "step": 928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5876887193641183e-05,
+      "loss": 1.7095,
+      "step": 929
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5867229301482227e-05,
+      "loss": 1.8886,
+      "step": 930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.585756305654338e-05,
+      "loss": 1.395,
+      "step": 931
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5847888472585826e-05,
+      "loss": 1.7254,
+      "step": 932
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.583820556338262e-05,
+      "loss": 1.704,
+      "step": 933
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.582851434271867e-05,
+      "loss": 1.7093,
+      "step": 934
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5818814824390713e-05,
+      "loss": 1.7495,
+      "step": 935
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.58091070222073e-05,
+      "loss": 1.9524,
+      "step": 936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5799390949988775e-05,
+      "loss": 1.7421,
+      "step": 937
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.578966662156726e-05,
+      "loss": 1.6707,
+      "step": 938
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5779934050786633e-05,
+      "loss": 1.5381,
+      "step": 939
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.577019325150249e-05,
+      "loss": 1.4886,
+      "step": 940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5760444237582157e-05,
+      "loss": 1.7248,
+      "step": 941
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5750687022904654e-05,
+      "loss": 1.4004,
+      "step": 942
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5740921621360664e-05,
+      "loss": 1.7162,
+      "step": 943
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5731148046852537e-05,
+      "loss": 1.6786,
+      "step": 944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.572136631329425e-05,
+      "loss": 1.442,
+      "step": 945
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5711576434611404e-05,
+      "loss": 1.5372,
+      "step": 946
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5701778424741188e-05,
+      "loss": 1.4451,
+      "step": 947
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5691972297632373e-05,
+      "loss": 1.5666,
+      "step": 948
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.568215806724528e-05,
+      "loss": 1.7347,
+      "step": 949
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5672335747551772e-05,
+      "loss": 1.4896,
+      "step": 950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.566250535253522e-05,
+      "loss": 1.57,
+      "step": 951
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5652666896190498e-05,
+      "loss": 1.6703,
+      "step": 952
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5642820392523962e-05,
+      "loss": 1.4305,
+      "step": 953
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.563296585555341e-05,
+      "loss": 1.6613,
+      "step": 954
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5623103299308082e-05,
+      "loss": 1.9669,
+      "step": 955
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5613232737828644e-05,
+      "loss": 1.7696,
+      "step": 956
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.560335418516714e-05,
+      "loss": 1.6591,
+      "step": 957
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5593467655387012e-05,
+      "loss": 1.5471,
+      "step": 958
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5583573162563045e-05,
+      "loss": 1.3704,
+      "step": 959
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5573670720781363e-05,
+      "loss": 1.6398,
+      "step": 960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.55637603441394e-05,
+      "loss": 1.7751,
+      "step": 961
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5553842046745903e-05,
+      "loss": 1.4721,
+      "step": 962
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.554391584272088e-05,
+      "loss": 1.5168,
+      "step": 963
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5533981746195598e-05,
+      "loss": 1.4701,
+      "step": 964
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5524039771312565e-05,
+      "loss": 1.5164,
+      "step": 965
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5514089932225506e-05,
+      "loss": 1.6643,
+      "step": 966
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5504132243099327e-05,
+      "loss": 1.5087,
+      "step": 967
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5494166718110137e-05,
+      "loss": 1.6638,
+      "step": 968
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5484193371445168e-05,
+      "loss": 1.4278,
+      "step": 969
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5474212217302814e-05,
+      "loss": 1.5084,
+      "step": 970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5464223269892564e-05,
+      "loss": 1.4425,
+      "step": 971
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.545422654343502e-05,
+      "loss": 1.7087,
+      "step": 972
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.544422205216184e-05,
+      "loss": 1.4237,
+      "step": 973
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.543420981031576e-05,
+      "loss": 1.7358,
+      "step": 974
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5424189832150518e-05,
+      "loss": 1.8258,
+      "step": 975
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5414162131930896e-05,
+      "loss": 1.7415,
+      "step": 976
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5404126723932648e-05,
+      "loss": 1.8253,
+      "step": 977
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5394083622442516e-05,
+      "loss": 1.4271,
+      "step": 978
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5384032841758186e-05,
+      "loss": 1.6272,
+      "step": 979
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.537397439618828e-05,
+      "loss": 1.6815,
+      "step": 980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.536390830005233e-05,
+      "loss": 1.6434,
+      "step": 981
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5353834567680758e-05,
+      "loss": 1.6524,
+      "step": 982
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5343753213414862e-05,
+      "loss": 1.6558,
+      "step": 983
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5333664251606787e-05,
+      "loss": 1.5875,
+      "step": 984
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5323567696619513e-05,
+      "loss": 1.527,
+      "step": 985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.531346356282682e-05,
+      "loss": 1.5793,
+      "step": 986
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.530335186461329e-05,
+      "loss": 1.6894,
+      "step": 987
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5293232616374267e-05,
+      "loss": 1.687,
+      "step": 988
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5283105832515842e-05,
+      "loss": 1.7821,
+      "step": 989
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5272971527454838e-05,
+      "loss": 1.8022,
+      "step": 990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5262829715618782e-05,
+      "loss": 1.586,
+      "step": 991
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5252680411445892e-05,
+      "loss": 1.6101,
+      "step": 992
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5242523629385048e-05,
+      "loss": 1.6305,
+      "step": 993
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5232359383895779e-05,
+      "loss": 1.7602,
+      "step": 994
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5222187689448235e-05,
+      "loss": 1.3457,
+      "step": 995
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.521200856052318e-05,
+      "loss": 1.6397,
+      "step": 996
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.520182201161195e-05,
+      "loss": 1.5046,
+      "step": 997
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5191628057216452e-05,
+      "loss": 1.6075,
+      "step": 998
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5181426711849133e-05,
+      "loss": 1.5093,
+      "step": 999
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.517121799003296e-05,
+      "loss": 1.7927,
+      "step": 1000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5161001906301407e-05,
+      "loss": 1.7316,
+      "step": 1001
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5150778475198427e-05,
+      "loss": 1.7274,
+      "step": 1002
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5140547711278428e-05,
+      "loss": 1.2897,
+      "step": 1003
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5130309629106264e-05,
+      "loss": 1.5808,
+      "step": 1004
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.51200642432572e-05,
+      "loss": 1.7323,
+      "step": 1005
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5109811568316906e-05,
+      "loss": 1.6818,
+      "step": 1006
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5099551618881426e-05,
+      "loss": 1.4173,
+      "step": 1007
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5089284409557158e-05,
+      "loss": 1.7259,
+      "step": 1008
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5079009954960842e-05,
+      "loss": 1.5218,
+      "step": 1009
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5068728269719524e-05,
+      "loss": 1.6281,
+      "step": 1010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.505843936847055e-05,
+      "loss": 1.7548,
+      "step": 1011
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5048143265861536e-05,
+      "loss": 1.524,
+      "step": 1012
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5037839976550352e-05,
+      "loss": 1.7357,
+      "step": 1013
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5027529515205097e-05,
+      "loss": 1.7994,
+      "step": 1014
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5017211896504082e-05,
+      "loss": 1.8074,
+      "step": 1015
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.500688713513581e-05,
+      "loss": 1.5332,
+      "step": 1016
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4996555245798944e-05,
+      "loss": 1.673,
+      "step": 1017
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4986216243202307e-05,
+      "loss": 1.6054,
+      "step": 1018
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4975870142064838e-05,
+      "loss": 1.8379,
+      "step": 1019
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4965516957115585e-05,
+      "loss": 1.7481,
+      "step": 1020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.495515670309368e-05,
+      "loss": 1.7256,
+      "step": 1021
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4944789394748322e-05,
+      "loss": 1.5541,
+      "step": 1022
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4934415046838755e-05,
+      "loss": 1.7022,
+      "step": 1023
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4924033674134236e-05,
+      "loss": 1.6277,
+      "step": 1024
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4913645291414027e-05,
+      "loss": 1.8026,
+      "step": 1025
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.490324991346737e-05,
+      "loss": 1.548,
+      "step": 1026
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4892847555093468e-05,
+      "loss": 1.7814,
+      "step": 1027
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4882438231101456e-05,
+      "loss": 1.3886,
+      "step": 1028
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4872021956310394e-05,
+      "loss": 1.503,
+      "step": 1029
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4861598745549228e-05,
+      "loss": 1.5944,
+      "step": 1030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4851168613656785e-05,
+      "loss": 1.5893,
+      "step": 1031
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.484073157548174e-05,
+      "loss": 1.4621,
+      "step": 1032
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4830287645882602e-05,
+      "loss": 1.6625,
+      "step": 1033
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.481983683972769e-05,
+      "loss": 1.6391,
+      "step": 1034
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4809379171895122e-05,
+      "loss": 1.7311,
+      "step": 1035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4798914657272771e-05,
+      "loss": 1.424,
+      "step": 1036
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4788443310758263e-05,
+      "loss": 1.5855,
+      "step": 1037
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.477796514725895e-05,
+      "loss": 1.5577,
+      "step": 1038
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4767480181691888e-05,
+      "loss": 1.469,
+      "step": 1039
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.475698842898382e-05,
+      "loss": 2.0158,
+      "step": 1040
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4746489904071148e-05,
+      "loss": 1.7775,
+      "step": 1041
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4735984621899917e-05,
+      "loss": 1.5493,
+      "step": 1042
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.472547259742579e-05,
+      "loss": 1.7657,
+      "step": 1043
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4714953845614028e-05,
+      "loss": 1.7562,
+      "step": 1044
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4704428381439471e-05,
+      "loss": 1.4397,
+      "step": 1045
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4693896219886518e-05,
+      "loss": 1.5493,
+      "step": 1046
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4683357375949099e-05,
+      "loss": 1.7458,
+      "step": 1047
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.467281186463065e-05,
+      "loss": 1.7092,
+      "step": 1048
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4662259700944117e-05,
+      "loss": 1.7902,
+      "step": 1049
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.46517008999119e-05,
+      "loss": 1.6822,
+      "step": 1050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4641135476565853e-05,
+      "loss": 1.4869,
+      "step": 1051
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4630563445947265e-05,
+      "loss": 1.2409,
+      "step": 1052
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4619984823106821e-05,
+      "loss": 1.6679,
+      "step": 1053
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4609399623104594e-05,
+      "loss": 1.4442,
+      "step": 1054
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4598807861010023e-05,
+      "loss": 1.7156,
+      "step": 1055
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4588209551901886e-05,
+      "loss": 1.8078,
+      "step": 1056
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4577604710868288e-05,
+      "loss": 1.7926,
+      "step": 1057
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4566993353006622e-05,
+      "loss": 1.2587,
+      "step": 1058
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4556375493423572e-05,
+      "loss": 1.9043,
+      "step": 1059
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4545751147235063e-05,
+      "loss": 1.6593,
+      "step": 1060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4535120329566268e-05,
+      "loss": 1.9199,
+      "step": 1061
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4524483055551561e-05,
+      "loss": 1.39,
+      "step": 1062
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4513839340334523e-05,
+      "loss": 1.5897,
+      "step": 1063
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4503189199067891e-05,
+      "loss": 1.695,
+      "step": 1064
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4492532646913553e-05,
+      "loss": 1.5598,
+      "step": 1065
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.448186969904253e-05,
+      "loss": 1.6194,
+      "step": 1066
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.447120037063494e-05,
+      "loss": 1.4702,
+      "step": 1067
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.446052467687999e-05,
+      "loss": 1.673,
+      "step": 1068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4449842632975948e-05,
+      "loss": 1.6351,
+      "step": 1069
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4439154254130123e-05,
+      "loss": 1.6767,
+      "step": 1070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4428459555558841e-05,
+      "loss": 1.518,
+      "step": 1071
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4417758552487424e-05,
+      "loss": 1.6418,
+      "step": 1072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4407051260150167e-05,
+      "loss": 1.4267,
+      "step": 1073
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4396337693790329e-05,
+      "loss": 1.392,
+      "step": 1074
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4385617868660094e-05,
+      "loss": 1.6811,
+      "step": 1075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4374891800020549e-05,
+      "loss": 1.3297,
+      "step": 1076
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4364159503141684e-05,
+      "loss": 1.9199,
+      "step": 1077
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4353420993302346e-05,
+      "loss": 1.5111,
+      "step": 1078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4342676285790224e-05,
+      "loss": 1.3439,
+      "step": 1079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4331925395901847e-05,
+      "loss": 1.7453,
+      "step": 1080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4321168338942528e-05,
+      "loss": 1.867,
+      "step": 1081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4310405130226365e-05,
+      "loss": 1.537,
+      "step": 1082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4299635785076214e-05,
+      "loss": 1.4933,
+      "step": 1083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4288860318823673e-05,
+      "loss": 1.7592,
+      "step": 1084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4278078746809039e-05,
+      "loss": 1.4784,
+      "step": 1085
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4267291084381322e-05,
+      "loss": 1.5672,
+      "step": 1086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4256497346898186e-05,
+      "loss": 1.5483,
+      "step": 1087
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4245697549725951e-05,
+      "loss": 1.753,
+      "step": 1088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4234891708239562e-05,
+      "loss": 1.6669,
+      "step": 1089
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4224079837822566e-05,
+      "loss": 1.6799,
+      "step": 1090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4213261953867099e-05,
+      "loss": 1.629,
+      "step": 1091
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4202438071773856e-05,
+      "loss": 1.4544,
+      "step": 1092
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4191608206952069e-05,
+      "loss": 1.7615,
+      "step": 1093
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4180772374819489e-05,
+      "loss": 1.5733,
+      "step": 1094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.416993059080236e-05,
+      "loss": 1.4325,
+      "step": 1095
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4159082870335402e-05,
+      "loss": 1.5521,
+      "step": 1096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4148229228861782e-05,
+      "loss": 1.6471,
+      "step": 1097
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4137369681833106e-05,
+      "loss": 1.5429,
+      "step": 1098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4126504244709377e-05,
+      "loss": 1.7703,
+      "step": 1099
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4115632932958992e-05,
+      "loss": 1.5304,
+      "step": 1100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4104755762058701e-05,
+      "loss": 1.6547,
+      "step": 1101
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.40938727474936e-05,
+      "loss": 1.7368,
+      "step": 1102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.408298390475711e-05,
+      "loss": 1.4461,
+      "step": 1103
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4072089249350942e-05,
+      "loss": 1.7066,
+      "step": 1104
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4061188796785085e-05,
+      "loss": 1.745,
+      "step": 1105
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4050282562577782e-05,
+      "loss": 1.6486,
+      "step": 1106
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4039370562255501e-05,
+      "loss": 1.4831,
+      "step": 1107
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4028452811352926e-05,
+      "loss": 1.54,
+      "step": 1108
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4017529325412926e-05,
+      "loss": 1.6582,
+      "step": 1109
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.400660011998653e-05,
+      "loss": 1.7603,
+      "step": 1110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3995665210632918e-05,
+      "loss": 1.6216,
+      "step": 1111
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.398472461291938e-05,
+      "loss": 1.6027,
+      "step": 1112
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3973778342421314e-05,
+      "loss": 1.6572,
+      "step": 1113
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3962826414722185e-05,
+      "loss": 1.4402,
+      "step": 1114
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.395186884541352e-05,
+      "loss": 1.6595,
+      "step": 1115
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3940905650094874e-05,
+      "loss": 1.5473,
+      "step": 1116
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3929936844373806e-05,
+      "loss": 1.653,
+      "step": 1117
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.391896244386587e-05,
+      "loss": 1.4939,
+      "step": 1118
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3907982464194584e-05,
+      "loss": 1.4917,
+      "step": 1119
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3896996920991409e-05,
+      "loss": 1.3911,
+      "step": 1120
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3886005829895717e-05,
+      "loss": 1.6069,
+      "step": 1121
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.387500920655479e-05,
+      "loss": 1.683,
+      "step": 1122
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3864007066623782e-05,
+      "loss": 1.9023,
+      "step": 1123
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3852999425765697e-05,
+      "loss": 1.5886,
+      "step": 1124
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.384198629965137e-05,
+      "loss": 1.5902,
+      "step": 1125
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3830967703959458e-05,
+      "loss": 1.7075,
+      "step": 1126
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3819943654376393e-05,
+      "loss": 1.8132,
+      "step": 1127
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3808914166596367e-05,
+      "loss": 1.7699,
+      "step": 1128
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3797879256321323e-05,
+      "loss": 1.8824,
+      "step": 1129
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.378683893926092e-05,
+      "loss": 1.7651,
+      "step": 1130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3775793231132515e-05,
+      "loss": 1.7531,
+      "step": 1131
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3764742147661143e-05,
+      "loss": 1.5328,
+      "step": 1132
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3753685704579489e-05,
+      "loss": 1.5323,
+      "step": 1133
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3742623917627864e-05,
+      "loss": 1.6862,
+      "step": 1134
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.373155680255419e-05,
+      "loss": 1.6841,
+      "step": 1135
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3720484375113978e-05,
+      "loss": 1.5371,
+      "step": 1136
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3709406651070299e-05,
+      "loss": 1.6985,
+      "step": 1137
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3698323646193758e-05,
+      "loss": 1.6268,
+      "step": 1138
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3687235376262492e-05,
+      "loss": 1.5333,
+      "step": 1139
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3676141857062117e-05,
+      "loss": 1.7055,
+      "step": 1140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.366504310438574e-05,
+      "loss": 1.4433,
+      "step": 1141
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3653939134033897e-05,
+      "loss": 1.8203,
+      "step": 1142
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3642829961814577e-05,
+      "loss": 1.6585,
+      "step": 1143
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3631715603543153e-05,
+      "loss": 1.4763,
+      "step": 1144
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.362059607504239e-05,
+      "loss": 1.8157,
+      "step": 1145
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3609471392142419e-05,
+      "loss": 1.5029,
+      "step": 1146
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3598341570680698e-05,
+      "loss": 1.5438,
+      "step": 1147
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3587206626502004e-05,
+      "loss": 1.4743,
+      "step": 1148
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3576066575458415e-05,
+      "loss": 1.8572,
+      "step": 1149
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3564921433409268e-05,
+      "loss": 1.5078,
+      "step": 1150
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3553771216221155e-05,
+      "loss": 1.2459,
+      "step": 1151
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3542615939767882e-05,
+      "loss": 1.7684,
+      "step": 1152
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3531455619930481e-05,
+      "loss": 1.4721,
+      "step": 1153
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3520290272597135e-05,
+      "loss": 1.4634,
+      "step": 1154
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3509119913663206e-05,
+      "loss": 1.5922,
+      "step": 1155
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3497944559031185e-05,
+      "loss": 1.58,
+      "step": 1156
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3486764224610667e-05,
+      "loss": 1.8095,
+      "step": 1157
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3475578926318343e-05,
+      "loss": 1.6782,
+      "step": 1158
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3464388680077973e-05,
+      "loss": 1.4844,
+      "step": 1159
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.345319350182036e-05,
+      "loss": 1.3102,
+      "step": 1160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3441993407483321e-05,
+      "loss": 1.719,
+      "step": 1161
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.343078841301168e-05,
+      "loss": 1.6825,
+      "step": 1162
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3419578534357236e-05,
+      "loss": 1.7969,
+      "step": 1163
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3408363787478736e-05,
+      "loss": 1.7454,
+      "step": 1164
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3397144188341865e-05,
+      "loss": 1.7881,
+      "step": 1165
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3385919752919208e-05,
+      "loss": 1.8708,
+      "step": 1166
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3374690497190244e-05,
+      "loss": 1.6696,
+      "step": 1167
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3363456437141305e-05,
+      "loss": 1.6845,
+      "step": 1168
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.335221758876557e-05,
+      "loss": 1.7927,
+      "step": 1169
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.334097396806303e-05,
+      "loss": 1.5664,
+      "step": 1170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.332972559104047e-05,
+      "loss": 1.7425,
+      "step": 1171
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3318472473711453e-05,
+      "loss": 1.5114,
+      "step": 1172
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3307214632096282e-05,
+      "loss": 1.4314,
+      "step": 1173
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.329595208222199e-05,
+      "loss": 1.7526,
+      "step": 1174
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3284684840122313e-05,
+      "loss": 1.693,
+      "step": 1175
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3273412921837663e-05,
+      "loss": 1.6599,
+      "step": 1176
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3262136343415117e-05,
+      "loss": 1.7349,
+      "step": 1177
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3250855120908379e-05,
+      "loss": 1.6321,
+      "step": 1178
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3239569270377768e-05,
+      "loss": 1.6261,
+      "step": 1179
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3228278807890184e-05,
+      "loss": 1.5685,
+      "step": 1180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3216983749519112e-05,
+      "loss": 1.6259,
+      "step": 1181
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3205684111344557e-05,
+      "loss": 1.7537,
+      "step": 1182
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.319437990945306e-05,
+      "loss": 1.6813,
+      "step": 1183
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3183071159937649e-05,
+      "loss": 1.5848,
+      "step": 1184
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3171757878897831e-05,
+      "loss": 1.9219,
+      "step": 1185
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3160440082439565e-05,
+      "loss": 1.6489,
+      "step": 1186
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3149117786675239e-05,
+      "loss": 1.7062,
+      "step": 1187
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3137791007723636e-05,
+      "loss": 1.5602,
+      "step": 1188
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3126459761709943e-05,
+      "loss": 1.7758,
+      "step": 1189
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.311512406476568e-05,
+      "loss": 1.4631,
+      "step": 1190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3103783933028726e-05,
+      "loss": 1.5428,
+      "step": 1191
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.309243938264326e-05,
+      "loss": 1.7362,
+      "step": 1192
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3081090429759755e-05,
+      "loss": 1.6797,
+      "step": 1193
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3069737090534951e-05,
+      "loss": 1.748,
+      "step": 1194
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.305837938113184e-05,
+      "loss": 1.6353,
+      "step": 1195
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3047017317719623e-05,
+      "loss": 1.533,
+      "step": 1196
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.303565091647371e-05,
+      "loss": 1.6425,
+      "step": 1197
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.302428019357568e-05,
+      "loss": 1.7006,
+      "step": 1198
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3012905165213265e-05,
+      "loss": 1.7035,
+      "step": 1199
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.300152584758033e-05,
+      "loss": 1.4196,
+      "step": 1200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2990142256876845e-05,
+      "loss": 1.6645,
+      "step": 1201
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2978754409308864e-05,
+      "loss": 1.687,
+      "step": 1202
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.29673623210885e-05,
+      "loss": 1.5081,
+      "step": 1203
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2955966008433898e-05,
+      "loss": 1.7733,
+      "step": 1204
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2944565487569224e-05,
+      "loss": 1.5592,
+      "step": 1205
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.293316077472464e-05,
+      "loss": 1.541,
+      "step": 1206
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.292175188613626e-05,
+      "loss": 1.5144,
+      "step": 1207
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2910338838046154e-05,
+      "loss": 1.4817,
+      "step": 1208
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2898921646702317e-05,
+      "loss": 1.5223,
+      "step": 1209
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2887500328358627e-05,
+      "loss": 1.6485,
+      "step": 1210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2876074899274855e-05,
+      "loss": 1.6818,
+      "step": 1211
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2864645375716614e-05,
+      "loss": 1.2953,
+      "step": 1212
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2853211773955347e-05,
+      "loss": 1.4401,
+      "step": 1213
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2841774110268304e-05,
+      "loss": 1.5226,
+      "step": 1214
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2830332400938518e-05,
+      "loss": 1.7461,
+      "step": 1215
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2818886662254782e-05,
+      "loss": 1.659,
+      "step": 1216
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2807436910511626e-05,
+      "loss": 1.6772,
+      "step": 1217
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.279598316200929e-05,
+      "loss": 1.7157,
+      "step": 1218
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2784525433053707e-05,
+      "loss": 1.5594,
+      "step": 1219
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2773063739956473e-05,
+      "loss": 1.4783,
+      "step": 1220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.276159809903483e-05,
+      "loss": 1.5933,
+      "step": 1221
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2750128526611642e-05,
+      "loss": 1.5352,
+      "step": 1222
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2738655039015368e-05,
+      "loss": 1.5875,
+      "step": 1223
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2727177652580044e-05,
+      "loss": 1.6181,
+      "step": 1224
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2715696383645247e-05,
+      "loss": 1.3225,
+      "step": 1225
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2704211248556089e-05,
+      "loss": 1.7181,
+      "step": 1226
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2692722263663186e-05,
+      "loss": 1.8566,
+      "step": 1227
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2681229445322633e-05,
+      "loss": 1.712,
+      "step": 1228
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2669732809895986e-05,
+      "loss": 1.7477,
+      "step": 1229
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.265823237375023e-05,
+      "loss": 1.6415,
+      "step": 1230
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2646728153257761e-05,
+      "loss": 1.6903,
+      "step": 1231
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2635220164796364e-05,
+      "loss": 1.6767,
+      "step": 1232
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.262370842474919e-05,
+      "loss": 1.557,
+      "step": 1233
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.261219294950473e-05,
+      "loss": 1.5915,
+      "step": 1234
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2600673755456789e-05,
+      "loss": 1.7127,
+      "step": 1235
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2589150859004473e-05,
+      "loss": 1.4802,
+      "step": 1236
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2577624276552155e-05,
+      "loss": 1.6932,
+      "step": 1237
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2566094024509452e-05,
+      "loss": 1.5752,
+      "step": 1238
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2554560119291206e-05,
+      "loss": 1.5841,
+      "step": 1239
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2543022577317472e-05,
+      "loss": 1.739,
+      "step": 1240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2531481415013464e-05,
+      "loss": 1.7102,
+      "step": 1241
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2519936648809564e-05,
+      "loss": 1.7832,
+      "step": 1242
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2508388295141276e-05,
+      "loss": 1.8557,
+      "step": 1243
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.249683637044922e-05,
+      "loss": 1.3856,
+      "step": 1244
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2485280891179086e-05,
+      "loss": 1.6422,
+      "step": 1245
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.247372187378164e-05,
+      "loss": 2.1887,
+      "step": 1246
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2462159334712676e-05,
+      "loss": 1.9006,
+      "step": 1247
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2450593290433005e-05,
+      "loss": 1.7601,
+      "step": 1248
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.243902375740842e-05,
+      "loss": 1.734,
+      "step": 1249
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2427450752109696e-05,
+      "loss": 1.4335,
+      "step": 1250
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2415874291012538e-05,
+      "loss": 1.809,
+      "step": 1251
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2404294390597576e-05,
+      "loss": 1.406,
+      "step": 1252
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2392711067350337e-05,
+      "loss": 1.5234,
+      "step": 1253
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2381124337761216e-05,
+      "loss": 1.7456,
+      "step": 1254
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2369534218325465e-05,
+      "loss": 1.8913,
+      "step": 1255
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2357940725543156e-05,
+      "loss": 1.6218,
+      "step": 1256
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2346343875919163e-05,
+      "loss": 1.6535,
+      "step": 1257
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2334743685963146e-05,
+      "loss": 1.4896,
+      "step": 1258
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2323140172189515e-05,
+      "loss": 1.3186,
+      "step": 1259
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2311533351117406e-05,
+      "loss": 1.5025,
+      "step": 1260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2299923239270675e-05,
+      "loss": 1.5083,
+      "step": 1261
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2288309853177855e-05,
+      "loss": 1.4921,
+      "step": 1262
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.227669320937215e-05,
+      "loss": 1.65,
+      "step": 1263
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2265073324391388e-05,
+      "loss": 1.6036,
+      "step": 1264
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.225345021477802e-05,
+      "loss": 1.4732,
+      "step": 1265
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2241823897079084e-05,
+      "loss": 1.6494,
+      "step": 1266
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2230194387846189e-05,
+      "loss": 1.7075,
+      "step": 1267
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2218561703635484e-05,
+      "loss": 1.7934,
+      "step": 1268
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2206925861007639e-05,
+      "loss": 1.818,
+      "step": 1269
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2195286876527824e-05,
+      "loss": 1.7087,
+      "step": 1270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2183644766765674e-05,
+      "loss": 1.5402,
+      "step": 1271
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2171999548295284e-05,
+      "loss": 1.6554,
+      "step": 1272
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2160351237695162e-05,
+      "loss": 1.5047,
+      "step": 1273
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.214869985154823e-05,
+      "loss": 1.5085,
+      "step": 1274
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.213704540644178e-05,
+      "loss": 1.5607,
+      "step": 1275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2125387918967461e-05,
+      "loss": 1.4395,
+      "step": 1276
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2113727405721261e-05,
+      "loss": 1.5419,
+      "step": 1277
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2102063883303462e-05,
+      "loss": 1.6719,
+      "step": 1278
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2090397368318634e-05,
+      "loss": 1.4981,
+      "step": 1279
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.207872787737562e-05,
+      "loss": 1.7986,
+      "step": 1280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2067055427087482e-05,
+      "loss": 1.6052,
+      "step": 1281
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2055380034071504e-05,
+      "loss": 1.4215,
+      "step": 1282
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.204370171494916e-05,
+      "loss": 1.5639,
+      "step": 1283
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2032020486346085e-05,
+      "loss": 1.6687,
+      "step": 1284
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2020336364892062e-05,
+      "loss": 1.7765,
+      "step": 1285
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2008649367220988e-05,
+      "loss": 1.7461,
+      "step": 1286
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1996959509970857e-05,
+      "loss": 1.542,
+      "step": 1287
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.198526680978373e-05,
+      "loss": 1.7398,
+      "step": 1288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.197357128330572e-05,
+      "loss": 1.5469,
+      "step": 1289
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1961872947186958e-05,
+      "loss": 1.5236,
+      "step": 1290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1950171818081583e-05,
+      "loss": 1.5148,
+      "step": 1291
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1938467912647708e-05,
+      "loss": 1.7556,
+      "step": 1292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1926761247547392e-05,
+      "loss": 1.5149,
+      "step": 1293
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1915051839446627e-05,
+      "loss": 1.6909,
+      "step": 1294
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.190333970501531e-05,
+      "loss": 1.7696,
+      "step": 1295
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1891624860927222e-05,
+      "loss": 1.7218,
+      "step": 1296
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1879907323859997e-05,
+      "loss": 1.5859,
+      "step": 1297
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1868187110495104e-05,
+      "loss": 1.8859,
+      "step": 1298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1856464237517826e-05,
+      "loss": 1.793,
+      "step": 1299
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1844738721617228e-05,
+      "loss": 1.657,
+      "step": 1300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1833010579486135e-05,
+      "loss": 1.6519,
+      "step": 1301
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1821279827821118e-05,
+      "loss": 1.5418,
+      "step": 1302
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1809546483322458e-05,
+      "loss": 1.7052,
+      "step": 1303
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1797810562694127e-05,
+      "loss": 1.5585,
+      "step": 1304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1786072082643774e-05,
+      "loss": 1.7492,
+      "step": 1305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1774331059882676e-05,
+      "loss": 1.5275,
+      "step": 1306
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1762587511125738e-05,
+      "loss": 1.7395,
+      "step": 1307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1750841453091464e-05,
+      "loss": 1.4839,
+      "step": 1308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1739092902501927e-05,
+      "loss": 1.5222,
+      "step": 1309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1727341876082748e-05,
+      "loss": 1.5207,
+      "step": 1310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1715588390563069e-05,
+      "loss": 1.2958,
+      "step": 1311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1703832462675544e-05,
+      "loss": 1.8622,
+      "step": 1312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1692074109156292e-05,
+      "loss": 1.8553,
+      "step": 1313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1680313346744897e-05,
+      "loss": 1.7046,
+      "step": 1314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1668550192184358e-05,
+      "loss": 1.5634,
+      "step": 1315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.165678466222109e-05,
+      "loss": 1.4167,
+      "step": 1316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.164501677360489e-05,
+      "loss": 1.4836,
+      "step": 1317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1633246543088903e-05,
+      "loss": 1.4215,
+      "step": 1318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.162147398742962e-05,
+      "loss": 1.4499,
+      "step": 1319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.160969912338684e-05,
+      "loss": 1.6927,
+      "step": 1320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1597921967723638e-05,
+      "loss": 1.6764,
+      "step": 1321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1586142537206365e-05,
+      "loss": 1.3818,
+      "step": 1322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.15743608486046e-05,
+      "loss": 1.5537,
+      "step": 1323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1562576918691141e-05,
+      "loss": 1.7134,
+      "step": 1324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1550790764241979e-05,
+      "loss": 1.2854,
+      "step": 1325
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.153900240203627e-05,
+      "loss": 1.4848,
+      "step": 1326
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.152721184885631e-05,
+      "loss": 1.448,
+      "step": 1327
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1515419121487522e-05,
+      "loss": 1.7475,
+      "step": 1328
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.150362423671841e-05,
+      "loss": 1.4458,
+      "step": 1329
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1491827211340568e-05,
+      "loss": 1.3791,
+      "step": 1330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1480028062148622e-05,
+      "loss": 1.6132,
+      "step": 1331
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1468226805940227e-05,
+      "loss": 1.4072,
+      "step": 1332
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1456423459516047e-05,
+      "loss": 1.5475,
+      "step": 1333
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1444618039679702e-05,
+      "loss": 1.5749,
+      "step": 1334
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.143281056323778e-05,
+      "loss": 1.7015,
+      "step": 1335
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1421001046999787e-05,
+      "loss": 1.4643,
+      "step": 1336
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1409189507778143e-05,
+      "loss": 1.5289,
+      "step": 1337
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1397375962388137e-05,
+      "loss": 1.7965,
+      "step": 1338
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1385560427647923e-05,
+      "loss": 1.617,
+      "step": 1339
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1373742920378483e-05,
+      "loss": 1.5381,
+      "step": 1340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1361923457403607e-05,
+      "loss": 1.6884,
+      "step": 1341
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1350102055549868e-05,
+      "loss": 1.4769,
+      "step": 1342
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1338278731646603e-05,
+      "loss": 1.6484,
+      "step": 1343
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1326453502525886e-05,
+      "loss": 1.5862,
+      "step": 1344
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1314626385022493e-05,
+      "loss": 1.4017,
+      "step": 1345
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1302797395973906e-05,
+      "loss": 1.5226,
+      "step": 1346
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1290966552220253e-05,
+      "loss": 1.5853,
+      "step": 1347
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1279133870604313e-05,
+      "loss": 1.6298,
+      "step": 1348
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1267299367971482e-05,
+      "loss": 1.7236,
+      "step": 1349
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1255463061169744e-05,
+      "loss": 1.5666,
+      "step": 1350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1243624967049653e-05,
+      "loss": 1.6452,
+      "step": 1351
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1231785102464307e-05,
+      "loss": 1.5262,
+      "step": 1352
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1219943484269329e-05,
+      "loss": 1.8981,
+      "step": 1353
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1208100129322827e-05,
+      "loss": 1.8015,
+      "step": 1354
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1196255054485396e-05,
+      "loss": 1.7094,
+      "step": 1355
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1184408276620074e-05,
+      "loss": 1.8474,
+      "step": 1356
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1172559812592316e-05,
+      "loss": 1.4666,
+      "step": 1357
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1160709679269986e-05,
+      "loss": 1.5856,
+      "step": 1358
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.114885789352332e-05,
+      "loss": 1.7492,
+      "step": 1359
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1137004472224915e-05,
+      "loss": 1.6024,
+      "step": 1360
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1125149432249687e-05,
+      "loss": 1.6389,
+      "step": 1361
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1113292790474857e-05,
+      "loss": 1.4293,
+      "step": 1362
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1101434563779929e-05,
+      "loss": 1.8504,
+      "step": 1363
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1089574769046665e-05,
+      "loss": 1.6104,
+      "step": 1364
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1077713423159056e-05,
+      "loss": 1.5039,
+      "step": 1365
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1065850543003302e-05,
+      "loss": 1.5566,
+      "step": 1366
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1053986145467795e-05,
+      "loss": 1.4938,
+      "step": 1367
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.104212024744307e-05,
+      "loss": 1.6359,
+      "step": 1368
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1030252865821814e-05,
+      "loss": 1.5906,
+      "step": 1369
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1018384017498817e-05,
+      "loss": 1.6035,
+      "step": 1370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1006513719370963e-05,
+      "loss": 1.7358,
+      "step": 1371
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.09946419883372e-05,
+      "loss": 1.7031,
+      "step": 1372
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0982768841298505e-05,
+      "loss": 1.6675,
+      "step": 1373
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0970894295157887e-05,
+      "loss": 1.7351,
+      "step": 1374
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0959018366820337e-05,
+      "loss": 1.5233,
+      "step": 1375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0947141073192813e-05,
+      "loss": 1.3807,
+      "step": 1376
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0935262431184224e-05,
+      "loss": 1.6437,
+      "step": 1377
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.092338245770539e-05,
+      "loss": 1.5475,
+      "step": 1378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0911501169669032e-05,
+      "loss": 1.7788,
+      "step": 1379
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.089961858398974e-05,
+      "loss": 1.6773,
+      "step": 1380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0887734717583953e-05,
+      "loss": 1.4647,
+      "step": 1381
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0875849587369931e-05,
+      "loss": 1.4252,
+      "step": 1382
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.086396321026774e-05,
+      "loss": 1.5415,
+      "step": 1383
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0852075603199213e-05,
+      "loss": 1.4866,
+      "step": 1384
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0840186783087936e-05,
+      "loss": 1.3765,
+      "step": 1385
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0828296766859224e-05,
+      "loss": 1.6217,
+      "step": 1386
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0816405571440089e-05,
+      "loss": 1.6645,
+      "step": 1387
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0804513213759234e-05,
+      "loss": 1.3922,
+      "step": 1388
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0792619710747004e-05,
+      "loss": 1.5749,
+      "step": 1389
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0780725079335383e-05,
+      "loss": 1.5204,
+      "step": 1390
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0768829336457958e-05,
+      "loss": 1.2803,
+      "step": 1391
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0756932499049894e-05,
+      "loss": 1.6384,
+      "step": 1392
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.074503458404792e-05,
+      "loss": 1.4258,
+      "step": 1393
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.07331356083903e-05,
+      "loss": 1.4932,
+      "step": 1394
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0721235589016805e-05,
+      "loss": 1.5761,
+      "step": 1395
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0709334542868692e-05,
+      "loss": 1.8186,
+      "step": 1396
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0697432486888681e-05,
+      "loss": 1.5917,
+      "step": 1397
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.068552943802093e-05,
+      "loss": 1.7973,
+      "step": 1398
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0673625413211002e-05,
+      "loss": 1.5501,
+      "step": 1399
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0661720429405866e-05,
+      "loss": 1.5497,
+      "step": 1400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0649814503553844e-05,
+      "loss": 1.4764,
+      "step": 1401
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.06379076526046e-05,
+      "loss": 1.4711,
+      "step": 1402
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0625999893509122e-05,
+      "loss": 1.713,
+      "step": 1403
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0614091243219682e-05,
+      "loss": 1.5149,
+      "step": 1404
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0602181718689822e-05,
+      "loss": 1.4376,
+      "step": 1405
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0590271336874339e-05,
+      "loss": 1.6151,
+      "step": 1406
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0578360114729236e-05,
+      "loss": 1.4859,
+      "step": 1407
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0566448069211723e-05,
+      "loss": 1.3853,
+      "step": 1408
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0554535217280178e-05,
+      "loss": 1.7395,
+      "step": 1409
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0542621575894125e-05,
+      "loss": 1.4653,
+      "step": 1410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0530707162014218e-05,
+      "loss": 1.6122,
+      "step": 1411
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0518791992602204e-05,
+      "loss": 1.6813,
+      "step": 1412
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0506876084620911e-05,
+      "loss": 1.5162,
+      "step": 1413
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0494959455034215e-05,
+      "loss": 1.5368,
+      "step": 1414
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0483042120807016e-05,
+      "loss": 1.6324,
+      "step": 1415
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0471124098905228e-05,
+      "loss": 1.4531,
+      "step": 1416
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0459205406295737e-05,
+      "loss": 1.5531,
+      "step": 1417
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.044728605994638e-05,
+      "loss": 1.5739,
+      "step": 1418
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0435366076825933e-05,
+      "loss": 1.7043,
+      "step": 1419
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0423445473904072e-05,
+      "loss": 1.6949,
+      "step": 1420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0411524268151359e-05,
+      "loss": 1.5442,
+      "step": 1421
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.039960247653921e-05,
+      "loss": 1.642,
+      "step": 1422
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0387680116039884e-05,
+      "loss": 1.4562,
+      "step": 1423
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0375757203626445e-05,
+      "loss": 1.5614,
+      "step": 1424
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0363833756272734e-05,
+      "loss": 1.5004,
+      "step": 1425
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0351909790953367e-05,
+      "loss": 1.6549,
+      "step": 1426
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0339985324643688e-05,
+      "loss": 1.6174,
+      "step": 1427
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0328060374319759e-05,
+      "loss": 1.65,
+      "step": 1428
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0316134956958334e-05,
+      "loss": 1.7216,
+      "step": 1429
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0304209089536827e-05,
+      "loss": 1.6205,
+      "step": 1430
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0292282789033292e-05,
+      "loss": 1.6978,
+      "step": 1431
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0280356072426398e-05,
+      "loss": 1.6674,
+      "step": 1432
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0268428956695413e-05,
+      "loss": 1.5286,
+      "step": 1433
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0256501458820172e-05,
+      "loss": 1.6045,
+      "step": 1434
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0244573595781045e-05,
+      "loss": 1.5994,
+      "step": 1435
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0232645384558935e-05,
+      "loss": 1.6663,
+      "step": 1436
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0220716842135234e-05,
+      "loss": 1.4722,
+      "step": 1437
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0208787985491801e-05,
+      "loss": 1.5185,
+      "step": 1438
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0196858831610951e-05,
+      "loss": 1.4893,
+      "step": 1439
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.018492939747542e-05,
+      "loss": 1.5822,
+      "step": 1440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0172999700068338e-05,
+      "loss": 1.4596,
+      "step": 1441
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0161069756373212e-05,
+      "loss": 1.6666,
+      "step": 1442
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0149139583373906e-05,
+      "loss": 1.6678,
+      "step": 1443
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.01372091980546e-05,
+      "loss": 1.535,
+      "step": 1444
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0125278617399784e-05,
+      "loss": 1.491,
+      "step": 1445
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0113347858394223e-05,
+      "loss": 1.7168,
+      "step": 1446
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0101416938022936e-05,
+      "loss": 1.6208,
+      "step": 1447
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0089485873271176e-05,
+      "loss": 1.3264,
+      "step": 1448
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0077554681124391e-05,
+      "loss": 1.5177,
+      "step": 1449
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0065623378568223e-05,
+      "loss": 1.8232,
+      "step": 1450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.005369198258846e-05,
+      "loss": 1.5726,
+      "step": 1451
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0041760510171032e-05,
+      "loss": 1.8903,
+      "step": 1452
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0029828978301977e-05,
+      "loss": 1.6815,
+      "step": 1453
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0017897403967408e-05,
+      "loss": 1.6521,
+      "step": 1454
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0005965804153508e-05,
+      "loss": 1.4628,
+      "step": 1455
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.994034195846495e-06,
+      "loss": 1.4936,
+      "step": 1456
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.982102596032597e-06,
+      "loss": 1.6062,
+      "step": 1457
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.970171021698027e-06,
+      "loss": 1.5715,
+      "step": 1458
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.958239489828968e-06,
+      "loss": 1.3657,
+      "step": 1459
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.946308017411543e-06,
+      "loss": 1.6497,
+      "step": 1460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.93437662143178e-06,
+      "loss": 1.6405,
+      "step": 1461
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.922445318875612e-06,
+      "loss": 1.3632,
+      "step": 1462
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.910514126728827e-06,
+      "loss": 1.4618,
+      "step": 1463
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.898583061977068e-06,
+      "loss": 1.7279,
+      "step": 1464
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.88665214160578e-06,
+      "loss": 1.5738,
+      "step": 1465
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.874721382600218e-06,
+      "loss": 1.8553,
+      "step": 1466
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.862790801945403e-06,
+      "loss": 1.4768,
+      "step": 1467
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.850860416626096e-06,
+      "loss": 1.5988,
+      "step": 1468
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.838930243626791e-06,
+      "loss": 1.6183,
+      "step": 1469
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.827000299931666e-06,
+      "loss": 1.5499,
+      "step": 1470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.815070602524586e-06,
+      "loss": 1.5744,
+      "step": 1471
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.80314116838905e-06,
+      "loss": 1.4634,
+      "step": 1472
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.7912120145082e-06,
+      "loss": 1.4936,
+      "step": 1473
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.77928315786477e-06,
+      "loss": 1.7769,
+      "step": 1474
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.767354615441066e-06,
+      "loss": 1.626,
+      "step": 1475
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.755426404218958e-06,
+      "loss": 1.6154,
+      "step": 1476
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.743498541179832e-06,
+      "loss": 2.0258,
+      "step": 1477
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.731571043304589e-06,
+      "loss": 1.7951,
+      "step": 1478
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.719643927573604e-06,
+      "loss": 1.3674,
+      "step": 1479
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.70771721096671e-06,
+      "loss": 1.7213,
+      "step": 1480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.695790910463176e-06,
+      "loss": 1.8304,
+      "step": 1481
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.683865043041664e-06,
+      "loss": 1.5519,
+      "step": 1482
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.671939625680243e-06,
+      "loss": 1.6132,
+      "step": 1483
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.660014675356315e-06,
+      "loss": 1.5612,
+      "step": 1484
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.648090209046638e-06,
+      "loss": 1.5789,
+      "step": 1485
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.636166243727269e-06,
+      "loss": 1.5749,
+      "step": 1486
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.624242796373557e-06,
+      "loss": 1.641,
+      "step": 1487
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.612319883960117e-06,
+      "loss": 1.5902,
+      "step": 1488
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.60039752346079e-06,
+      "loss": 1.7535,
+      "step": 1489
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.588475731848645e-06,
+      "loss": 1.5461,
+      "step": 1490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.576554526095931e-06,
+      "loss": 1.8523,
+      "step": 1491
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.564633923174072e-06,
+      "loss": 1.739,
+      "step": 1492
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.552713940053622e-06,
+      "loss": 1.4078,
+      "step": 1493
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.540794593704265e-06,
+      "loss": 1.4988,
+      "step": 1494
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.528875901094775e-06,
+      "loss": 1.371,
+      "step": 1495
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.516957879192987e-06,
+      "loss": 1.742,
+      "step": 1496
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.50504054496579e-06,
+      "loss": 1.8256,
+      "step": 1497
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.49312391537909e-06,
+      "loss": 1.2345,
+      "step": 1498
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.4812080073978e-06,
+      "loss": 1.4356,
+      "step": 1499
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.469292837985786e-06,
+      "loss": 1.4514,
+      "step": 1500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.457378424105875e-06,
+      "loss": 1.6653,
+      "step": 1501
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.445464782719824e-06,
+      "loss": 1.4308,
+      "step": 1502
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.433551930788278e-06,
+      "loss": 1.5736,
+      "step": 1503
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.421639885270769e-06,
+      "loss": 1.6293,
+      "step": 1504
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.409728663125664e-06,
+      "loss": 1.529,
+      "step": 1505
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.397818281310182e-06,
+      "loss": 1.4393,
+      "step": 1506
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.385908756780323e-06,
+      "loss": 1.6974,
+      "step": 1507
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.374000106490882e-06,
+      "loss": 1.33,
+      "step": 1508
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.362092347395401e-06,
+      "loss": 1.3257,
+      "step": 1509
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.350185496446157e-06,
+      "loss": 1.6084,
+      "step": 1510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.338279570594137e-06,
+      "loss": 1.5949,
+      "step": 1511
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.326374586789e-06,
+      "loss": 1.4288,
+      "step": 1512
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.314470561979076e-06,
+      "loss": 1.6462,
+      "step": 1513
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.302567513111322e-06,
+      "loss": 1.6326,
+      "step": 1514
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.290665457131308e-06,
+      "loss": 1.5415,
+      "step": 1515
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.278764410983198e-06,
+      "loss": 1.5942,
+      "step": 1516
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.266864391609702e-06,
+      "loss": 1.489,
+      "step": 1517
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.254965415952083e-06,
+      "loss": 1.6327,
+      "step": 1518
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.243067500950109e-06,
+      "loss": 1.6492,
+      "step": 1519
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.231170663542048e-06,
+      "loss": 1.6279,
+      "step": 1520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.219274920664619e-06,
+      "loss": 1.6027,
+      "step": 1521
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.207380289252996e-06,
+      "loss": 1.8429,
+      "step": 1522
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.195486786240771e-06,
+      "loss": 1.5398,
+      "step": 1523
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.183594428559913e-06,
+      "loss": 1.6267,
+      "step": 1524
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.171703233140781e-06,
+      "loss": 1.7189,
+      "step": 1525
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.159813216912067e-06,
+      "loss": 1.758,
+      "step": 1526
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.147924396800792e-06,
+      "loss": 1.5452,
+      "step": 1527
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.136036789732261e-06,
+      "loss": 1.5664,
+      "step": 1528
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.124150412630069e-06,
+      "loss": 1.761,
+      "step": 1529
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.11226528241605e-06,
+      "loss": 1.4514,
+      "step": 1530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.100381416010262e-06,
+      "loss": 1.4175,
+      "step": 1531
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.088498830330974e-06,
+      "loss": 1.7212,
+      "step": 1532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.076617542294613e-06,
+      "loss": 1.5689,
+      "step": 1533
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.064737568815783e-06,
+      "loss": 1.6351,
+      "step": 1534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.052858926807189e-06,
+      "loss": 1.569,
+      "step": 1535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.040981633179666e-06,
+      "loss": 1.5106,
+      "step": 1536
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.029105704842114e-06,
+      "loss": 1.3715,
+      "step": 1537
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.017231158701494e-06,
+      "loss": 1.5932,
+      "step": 1538
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.005358011662805e-06,
+      "loss": 1.754,
+      "step": 1539
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.993486280629039e-06,
+      "loss": 1.7342,
+      "step": 1540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.981615982501186e-06,
+      "loss": 1.8086,
+      "step": 1541
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.96974713417819e-06,
+      "loss": 1.4102,
+      "step": 1542
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.95787975255693e-06,
+      "loss": 1.6458,
+      "step": 1543
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.94601385453221e-06,
+      "loss": 1.451,
+      "step": 1544
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.934149456996696e-06,
+      "loss": 1.5115,
+      "step": 1545
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.922286576840948e-06,
+      "loss": 1.7639,
+      "step": 1546
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.910425230953339e-06,
+      "loss": 1.4725,
+      "step": 1547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.898565436220076e-06,
+      "loss": 1.5253,
+      "step": 1548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.886707209525148e-06,
+      "loss": 1.666,
+      "step": 1549
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.874850567750315e-06,
+      "loss": 1.2424,
+      "step": 1550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.862995527775089e-06,
+      "loss": 1.428,
+      "step": 1551
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.85114210647668e-06,
+      "loss": 1.2813,
+      "step": 1552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.839290320730018e-06,
+      "loss": 1.4706,
+      "step": 1553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.827440187407687e-06,
+      "loss": 1.7242,
+      "step": 1554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.815591723379931e-06,
+      "loss": 1.5579,
+      "step": 1555
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.803744945514606e-06,
+      "loss": 1.551,
+      "step": 1556
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.791899870677174e-06,
+      "loss": 1.331,
+      "step": 1557
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.780056515730674e-06,
+      "loss": 1.7178,
+      "step": 1558
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.768214897535693e-06,
+      "loss": 1.5405,
+      "step": 1559
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.75637503295035e-06,
+      "loss": 1.6926,
+      "step": 1560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.744536938830257e-06,
+      "loss": 1.6215,
+      "step": 1561
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.732700632028523e-06,
+      "loss": 1.4487,
+      "step": 1562
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.720866129395689e-06,
+      "loss": 1.4904,
+      "step": 1563
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.709033447779749e-06,
+      "loss": 1.3806,
+      "step": 1564
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.697202604026099e-06,
+      "loss": 1.5946,
+      "step": 1565
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.685373614977507e-06,
+      "loss": 1.7393,
+      "step": 1566
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.673546497474119e-06,
+      "loss": 1.6652,
+      "step": 1567
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.661721268353398e-06,
+      "loss": 1.6665,
+      "step": 1568
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.649897944450134e-06,
+      "loss": 1.4041,
+      "step": 1569
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.638076542596396e-06,
+      "loss": 1.8183,
+      "step": 1570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.626257079621517e-06,
+      "loss": 1.4689,
+      "step": 1571
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.61443957235208e-06,
+      "loss": 1.4563,
+      "step": 1572
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.602624037611865e-06,
+      "loss": 1.8297,
+      "step": 1573
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.590810492221862e-06,
+      "loss": 1.4407,
+      "step": 1574
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.578998953000218e-06,
+      "loss": 1.5484,
+      "step": 1575
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.567189436762227e-06,
+      "loss": 1.57,
+      "step": 1576
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.555381960320301e-06,
+      "loss": 1.5272,
+      "step": 1577
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.543576540483955e-06,
+      "loss": 1.7057,
+      "step": 1578
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.531773194059774e-06,
+      "loss": 1.5611,
+      "step": 1579
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.51997193785138e-06,
+      "loss": 1.3945,
+      "step": 1580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.508172788659436e-06,
+      "loss": 1.7305,
+      "step": 1581
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.496375763281591e-06,
+      "loss": 1.6835,
+      "step": 1582
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.484580878512485e-06,
+      "loss": 1.5178,
+      "step": 1583
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.472788151143692e-06,
+      "loss": 1.6548,
+      "step": 1584
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.460997597963732e-06,
+      "loss": 1.6223,
+      "step": 1585
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.449209235758024e-06,
+      "loss": 1.441,
+      "step": 1586
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.437423081308862e-06,
+      "loss": 1.6071,
+      "step": 1587
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.425639151395406e-06,
+      "loss": 1.6885,
+      "step": 1588
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.413857462793637e-06,
+      "loss": 1.6291,
+      "step": 1589
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.402078032276367e-06,
+      "loss": 1.4595,
+      "step": 1590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.390300876613164e-06,
+      "loss": 1.7373,
+      "step": 1591
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.378526012570381e-06,
+      "loss": 1.5518,
+      "step": 1592
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.366753456911099e-06,
+      "loss": 1.7596,
+      "step": 1593
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.354983226395114e-06,
+      "loss": 1.4498,
+      "step": 1594
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.343215337778915e-06,
+      "loss": 1.4796,
+      "step": 1595
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.331449807815646e-06,
+      "loss": 1.5796,
+      "step": 1596
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.31968665325511e-06,
+      "loss": 1.438,
+      "step": 1597
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.307925890843711e-06,
+      "loss": 1.5338,
+      "step": 1598
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.296167537324459e-06,
+      "loss": 1.576,
+      "step": 1599
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.284411609436933e-06,
+      "loss": 1.6412,
+      "step": 1600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.272658123917256e-06,
+      "loss": 1.297,
+      "step": 1601
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.260907097498078e-06,
+      "loss": 1.865,
+      "step": 1602
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.24915854690854e-06,
+      "loss": 1.5555,
+      "step": 1603
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.237412488874266e-06,
+      "loss": 1.6192,
+      "step": 1604
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.225668940117327e-06,
+      "loss": 1.5302,
+      "step": 1605
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.213927917356228e-06,
+      "loss": 1.6969,
+      "step": 1606
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.202189437305874e-06,
+      "loss": 1.8516,
+      "step": 1607
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.190453516677545e-06,
+      "loss": 1.568,
+      "step": 1608
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.178720172178886e-06,
+      "loss": 1.7605,
+      "step": 1609
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.166989420513867e-06,
+      "loss": 1.2721,
+      "step": 1610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.155261278382773e-06,
+      "loss": 1.5424,
+      "step": 1611
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.143535762482176e-06,
+      "loss": 1.534,
+      "step": 1612
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.131812889504895e-06,
+      "loss": 1.3295,
+      "step": 1613
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.120092676140007e-06,
+      "loss": 1.4227,
+      "step": 1614
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.108375139072781e-06,
+      "loss": 1.6211,
+      "step": 1615
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.096660294984693e-06,
+      "loss": 1.6758,
+      "step": 1616
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.084948160553375e-06,
+      "loss": 1.7365,
+      "step": 1617
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.07323875245261e-06,
+      "loss": 1.7047,
+      "step": 1618
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.061532087352295e-06,
+      "loss": 1.3846,
+      "step": 1619
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.049828181918417e-06,
+      "loss": 1.5463,
+      "step": 1620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.038127052813044e-06,
+      "loss": 1.5289,
+      "step": 1621
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.026428716694284e-06,
+      "loss": 1.6241,
+      "step": 1622
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.014733190216275e-06,
+      "loss": 1.7027,
+      "step": 1623
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.003040490029145e-06,
+      "loss": 1.5098,
+      "step": 1624
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.991350632779012e-06,
+      "loss": 1.5048,
+      "step": 1625
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.979663635107941e-06,
+      "loss": 1.521,
+      "step": 1626
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.967979513653917e-06,
+      "loss": 1.6509,
+      "step": 1627
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.956298285050842e-06,
+      "loss": 1.6593,
+      "step": 1628
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.944619965928498e-06,
+      "loss": 1.6461,
+      "step": 1629
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.932944572912523e-06,
+      "loss": 1.6019,
+      "step": 1630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.921272122624384e-06,
+      "loss": 1.7219,
+      "step": 1631
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.909602631681366e-06,
+      "loss": 1.4021,
+      "step": 1632
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.897936116696542e-06,
+      "loss": 1.5456,
+      "step": 1633
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.88627259427874e-06,
+      "loss": 1.5395,
+      "step": 1634
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.87461208103254e-06,
+      "loss": 1.465,
+      "step": 1635
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.862954593558223e-06,
+      "loss": 1.6116,
+      "step": 1636
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.851300148451774e-06,
+      "loss": 1.7174,
+      "step": 1637
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.839648762304841e-06,
+      "loss": 1.5312,
+      "step": 1638
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.828000451704717e-06,
+      "loss": 1.6013,
+      "step": 1639
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.816355233234327e-06,
+      "loss": 1.5673,
+      "step": 1640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.804713123472178e-06,
+      "loss": 1.6745,
+      "step": 1641
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.793074138992364e-06,
+      "loss": 1.7497,
+      "step": 1642
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.78143829636452e-06,
+      "loss": 1.6192,
+      "step": 1643
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.769805612153816e-06,
+      "loss": 1.6124,
+      "step": 1644
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.758176102920918e-06,
+      "loss": 1.3206,
+      "step": 1645
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.746549785221982e-06,
+      "loss": 1.6474,
+      "step": 1646
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.734926675608616e-06,
+      "loss": 1.4634,
+      "step": 1647
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.723306790627852e-06,
+      "loss": 1.6819,
+      "step": 1648
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.711690146822147e-06,
+      "loss": 1.6936,
+      "step": 1649
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.700076760729328e-06,
+      "loss": 1.8941,
+      "step": 1650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.6884666488826e-06,
+      "loss": 1.381,
+      "step": 1651
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.67685982781049e-06,
+      "loss": 1.534,
+      "step": 1652
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.665256314036856e-06,
+      "loss": 1.6982,
+      "step": 1653
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.653656124080839e-06,
+      "loss": 1.5934,
+      "step": 1654
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.642059274456848e-06,
+      "loss": 1.7371,
+      "step": 1655
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.630465781674538e-06,
+      "loss": 1.4087,
+      "step": 1656
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.618875662238786e-06,
+      "loss": 1.612,
+      "step": 1657
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.607288932649669e-06,
+      "loss": 1.7681,
+      "step": 1658
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.595705609402427e-06,
+      "loss": 1.7609,
+      "step": 1659
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.584125708987464e-06,
+      "loss": 1.6934,
+      "step": 1660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5725492478903065e-06,
+      "loss": 1.4336,
+      "step": 1661
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.56097624259158e-06,
+      "loss": 1.712,
+      "step": 1662
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.549406709567001e-06,
+      "loss": 1.6162,
+      "step": 1663
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.537840665287327e-06,
+      "loss": 1.5464,
+      "step": 1664
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.526278126218365e-06,
+      "loss": 1.6153,
+      "step": 1665
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.514719108820918e-06,
+      "loss": 1.408,
+      "step": 1666
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5031636295507845e-06,
+      "loss": 1.5711,
+      "step": 1667
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.491611704858727e-06,
+      "loss": 1.4098,
+      "step": 1668
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.480063351190439e-06,
+      "loss": 1.3499,
+      "step": 1669
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.46851858498654e-06,
+      "loss": 1.4203,
+      "step": 1670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.456977422682532e-06,
+      "loss": 1.4782,
+      "step": 1671
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.445439880708796e-06,
+      "loss": 1.5853,
+      "step": 1672
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.433905975490552e-06,
+      "loss": 1.5873,
+      "step": 1673
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.422375723447846e-06,
+      "loss": 1.7037,
+      "step": 1674
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.410849140995528e-06,
+      "loss": 1.5418,
+      "step": 1675
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.399326244543211e-06,
+      "loss": 1.5074,
+      "step": 1676
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.387807050495275e-06,
+      "loss": 1.7242,
+      "step": 1677
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.376291575250812e-06,
+      "loss": 1.4759,
+      "step": 1678
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.3647798352036394e-06,
+      "loss": 1.3884,
+      "step": 1679
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.3532718467422415e-06,
+      "loss": 1.5181,
+      "step": 1680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.34176762624977e-06,
+      "loss": 1.7005,
+      "step": 1681
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.330267190104015e-06,
+      "loss": 1.431,
+      "step": 1682
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.318770554677366e-06,
+      "loss": 1.8979,
+      "step": 1683
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.307277736336815e-06,
+      "loss": 1.5267,
+      "step": 1684
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.295788751443913e-06,
+      "loss": 1.5903,
+      "step": 1685
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.284303616354759e-06,
+      "loss": 1.253,
+      "step": 1686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.27282234741996e-06,
+      "loss": 1.2221,
+      "step": 1687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.261344960984631e-06,
+      "loss": 1.7133,
+      "step": 1688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.249871473388361e-06,
+      "loss": 1.5898,
+      "step": 1689
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.238401900965172e-06,
+      "loss": 1.8137,
+      "step": 1690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.226936260043531e-06,
+      "loss": 1.3861,
+      "step": 1691
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.215474566946296e-06,
+      "loss": 1.7047,
+      "step": 1692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.204016837990715e-06,
+      "loss": 1.4074,
+      "step": 1693
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.192563089488377e-06,
+      "loss": 1.4883,
+      "step": 1694
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.181113337745218e-06,
+      "loss": 1.5436,
+      "step": 1695
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.169667599061484e-06,
+      "loss": 1.3418,
+      "step": 1696
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.158225889731698e-06,
+      "loss": 1.5738,
+      "step": 1697
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.146788226044658e-06,
+      "loss": 1.6208,
+      "step": 1698
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.1353546242833885e-06,
+      "loss": 1.2694,
+      "step": 1699
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.12392510072515e-06,
+      "loss": 1.5973,
+      "step": 1700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.112499671641375e-06,
+      "loss": 1.6895,
+      "step": 1701
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.1010783532976866e-06,
+      "loss": 1.3201,
+      "step": 1702
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.089661161953847e-06,
+      "loss": 1.6054,
+      "step": 1703
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.078248113863741e-06,
+      "loss": 1.6929,
+      "step": 1704
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.066839225275366e-06,
+      "loss": 1.347,
+      "step": 1705
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.0554345124307765e-06,
+      "loss": 1.6863,
+      "step": 1706
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.044033991566106e-06,
+      "loss": 1.659,
+      "step": 1707
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.032637678911504e-06,
+      "loss": 1.6838,
+      "step": 1708
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.021245590691136e-06,
+      "loss": 1.4575,
+      "step": 1709
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.009857743123156e-06,
+      "loss": 1.5452,
+      "step": 1710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.998474152419672e-06,
+      "loss": 1.5975,
+      "step": 1711
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.987094834786739e-06,
+      "loss": 1.6389,
+      "step": 1712
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.975719806424324e-06,
+      "loss": 1.7488,
+      "step": 1713
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.964349083526295e-06,
+      "loss": 1.6343,
+      "step": 1714
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.95298268228038e-06,
+      "loss": 1.8227,
+      "step": 1715
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.941620618868161e-06,
+      "loss": 1.8224,
+      "step": 1716
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.930262909465051e-06,
+      "loss": 1.1914,
+      "step": 1717
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.918909570240249e-06,
+      "loss": 1.7999,
+      "step": 1718
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.9075606173567435e-06,
+      "loss": 1.3275,
+      "step": 1719
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.8962160669712755e-06,
+      "loss": 1.6261,
+      "step": 1720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.884875935234323e-06,
+      "loss": 1.3259,
+      "step": 1721
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.8735402382900605e-06,
+      "loss": 1.5051,
+      "step": 1722
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.862208992276362e-06,
+      "loss": 1.447,
+      "step": 1723
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.850882213324766e-06,
+      "loss": 1.863,
+      "step": 1724
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.839559917560437e-06,
+      "loss": 1.5845,
+      "step": 1725
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.828242121102173e-06,
+      "loss": 1.5892,
+      "step": 1726
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.8169288400623555e-06,
+      "loss": 1.5102,
+      "step": 1727
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.805620090546947e-06,
+      "loss": 1.3536,
+      "step": 1728
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.794315888655446e-06,
+      "loss": 1.7994,
+      "step": 1729
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.783016250480891e-06,
+      "loss": 1.5345,
+      "step": 1730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.771721192109817e-06,
+      "loss": 1.7666,
+      "step": 1731
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.760430729622236e-06,
+      "loss": 1.5554,
+      "step": 1732
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.749144879091626e-06,
+      "loss": 1.6225,
+      "step": 1733
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.737863656584886e-06,
+      "loss": 1.8757,
+      "step": 1734
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.72658707816234e-06,
+      "loss": 1.6269,
+      "step": 1735
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.715315159877691e-06,
+      "loss": 1.7373,
+      "step": 1736
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.704047917778011e-06,
+      "loss": 1.5422,
+      "step": 1737
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.692785367903723e-06,
+      "loss": 1.4293,
+      "step": 1738
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.681527526288549e-06,
+      "loss": 1.6787,
+      "step": 1739
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.670274408959535e-06,
+      "loss": 1.5579,
+      "step": 1740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6590260319369745e-06,
+      "loss": 1.5948,
+      "step": 1741
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.647782411234436e-06,
+      "loss": 1.419,
+      "step": 1742
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.636543562858697e-06,
+      "loss": 1.7773,
+      "step": 1743
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.625309502809758e-06,
+      "loss": 1.464,
+      "step": 1744
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.614080247080794e-06,
+      "loss": 1.4038,
+      "step": 1745
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.602855811658138e-06,
+      "loss": 1.5296,
+      "step": 1746
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.591636212521266e-06,
+      "loss": 1.5991,
+      "step": 1747
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.580421465642767e-06,
+      "loss": 1.5339,
+      "step": 1748
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.569211586988324e-06,
+      "loss": 1.6199,
+      "step": 1749
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.558006592516683e-06,
+      "loss": 1.3909,
+      "step": 1750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.546806498179643e-06,
+      "loss": 1.5212,
+      "step": 1751
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.535611319922031e-06,
+      "loss": 1.6971,
+      "step": 1752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.524421073681659e-06,
+      "loss": 1.411,
+      "step": 1753
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.513235775389338e-06,
+      "loss": 1.5912,
+      "step": 1754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.502055440968819e-06,
+      "loss": 1.4404,
+      "step": 1755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.490880086336797e-06,
+      "loss": 1.4707,
+      "step": 1756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.479709727402868e-06,
+      "loss": 1.4696,
+      "step": 1757
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.468544380069523e-06,
+      "loss": 1.7753,
+      "step": 1758
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.457384060232119e-06,
+      "loss": 1.6027,
+      "step": 1759
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.4462287837788495e-06,
+      "loss": 1.5656,
+      "step": 1760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.435078566590738e-06,
+      "loss": 1.4584,
+      "step": 1761
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.423933424541588e-06,
+      "loss": 1.3464,
+      "step": 1762
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.412793373497999e-06,
+      "loss": 1.6881,
+      "step": 1763
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.401658429319305e-06,
+      "loss": 1.6128,
+      "step": 1764
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.390528607857582e-06,
+      "loss": 1.818,
+      "step": 1765
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.379403924957612e-06,
+      "loss": 1.859,
+      "step": 1766
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.36828439645685e-06,
+      "loss": 1.5989,
+      "step": 1767
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.357170038185428e-06,
+      "loss": 1.6319,
+      "step": 1768
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.346060865966104e-06,
+      "loss": 1.4426,
+      "step": 1769
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.3349568956142636e-06,
+      "loss": 1.8158,
+      "step": 1770
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.3238581429378846e-06,
+      "loss": 1.7956,
+      "step": 1771
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.312764623737511e-06,
+      "loss": 1.5887,
+      "step": 1772
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.301676353806245e-06,
+      "loss": 1.4651,
+      "step": 1773
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.2905933489297055e-06,
+      "loss": 1.4191,
+      "step": 1774
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.279515624886024e-06,
+      "loss": 1.4645,
+      "step": 1775
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.26844319744581e-06,
+      "loss": 1.6481,
+      "step": 1776
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.257376082372138e-06,
+      "loss": 1.5659,
+      "step": 1777
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.246314295420515e-06,
+      "loss": 1.3886,
+      "step": 1778
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.235257852338857e-06,
+      "loss": 1.4936,
+      "step": 1779
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.2242067688674876e-06,
+      "loss": 1.5716,
+      "step": 1780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.2131610607390835e-06,
+      "loss": 1.5199,
+      "step": 1781
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.202120743678682e-06,
+      "loss": 1.5851,
+      "step": 1782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.191085833403636e-06,
+      "loss": 1.567,
+      "step": 1783
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.180056345623608e-06,
+      "loss": 1.8849,
+      "step": 1784
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.169032296040542e-06,
+      "loss": 1.4365,
+      "step": 1785
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.158013700348628e-06,
+      "loss": 1.6608,
+      "step": 1786
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.1470005742343075e-06,
+      "loss": 1.5282,
+      "step": 1787
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.1359929333762206e-06,
+      "loss": 1.8344,
+      "step": 1788
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.124990793445214e-06,
+      "loss": 1.7696,
+      "step": 1789
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.113994170104285e-06,
+      "loss": 1.4385,
+      "step": 1790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.103003079008593e-06,
+      "loss": 1.5002,
+      "step": 1791
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0920175358054166e-06,
+      "loss": 1.7646,
+      "step": 1792
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.08103755613413e-06,
+      "loss": 1.5009,
+      "step": 1793
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.070063155626197e-06,
+      "loss": 1.6243,
+      "step": 1794
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.059094349905128e-06,
+      "loss": 1.5057,
+      "step": 1795
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.048131154586483e-06,
+      "loss": 1.3089,
+      "step": 1796
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.037173585277816e-06,
+      "loss": 1.7252,
+      "step": 1797
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.026221657578688e-06,
+      "loss": 1.8852,
+      "step": 1798
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.015275387080621e-06,
+      "loss": 1.7413,
+      "step": 1799
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.004334789367083e-06,
+      "loss": 1.6771,
+      "step": 1800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9933998800134726e-06,
+      "loss": 1.5189,
+      "step": 1801
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.982470674587078e-06,
+      "loss": 1.631,
+      "step": 1802
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.971547188647078e-06,
+      "loss": 1.4991,
+      "step": 1803
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9606294377445006e-06,
+      "loss": 1.5662,
+      "step": 1804
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.949717437422222e-06,
+      "loss": 1.6073,
+      "step": 1805
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.938811203214918e-06,
+      "loss": 1.8055,
+      "step": 1806
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.92791075064906e-06,
+      "loss": 1.7905,
+      "step": 1807
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.917016095242893e-06,
+      "loss": 1.5288,
+      "step": 1808
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9061272525064015e-06,
+      "loss": 1.9134,
+      "step": 1809
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.8952442379413045e-06,
+      "loss": 1.551,
+      "step": 1810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.88436706704101e-06,
+      "loss": 1.7406,
+      "step": 1811
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.873495755290621e-06,
+      "loss": 1.7211,
+      "step": 1812
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.862630318166896e-06,
+      "loss": 1.9039,
+      "step": 1813
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.851770771138218e-06,
+      "loss": 1.1796,
+      "step": 1814
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.840917129664602e-06,
+      "loss": 1.676,
+      "step": 1815
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.830069409197645e-06,
+      "loss": 1.4525,
+      "step": 1816
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.819227625180517e-06,
+      "loss": 1.6511,
+      "step": 1817
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.8083917930479365e-06,
+      "loss": 1.8143,
+      "step": 1818
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.797561928226145e-06,
+      "loss": 1.331,
+      "step": 1819
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.786738046132902e-06,
+      "loss": 1.7415,
+      "step": 1820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.775920162177437e-06,
+      "loss": 1.6579,
+      "step": 1821
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.765108291760443e-06,
+      "loss": 1.4423,
+      "step": 1822
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.754302450274053e-06,
+      "loss": 1.343,
+      "step": 1823
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.743502653101818e-06,
+      "loss": 1.4121,
+      "step": 1824
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.732708915618679e-06,
+      "loss": 1.5275,
+      "step": 1825
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.72192125319096e-06,
+      "loss": 1.8203,
+      "step": 1826
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.711139681176332e-06,
+      "loss": 1.4243,
+      "step": 1827
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.700364214923788e-06,
+      "loss": 1.4667,
+      "step": 1828
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.689594869773639e-06,
+      "loss": 1.7091,
+      "step": 1829
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.678831661057473e-06,
+      "loss": 1.4341,
+      "step": 1830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.668074604098157e-06,
+      "loss": 1.7594,
+      "step": 1831
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.657323714209775e-06,
+      "loss": 1.7115,
+      "step": 1832
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.646579006697659e-06,
+      "loss": 1.6253,
+      "step": 1833
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.63584049685832e-06,
+      "loss": 2.0217,
+      "step": 1834
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.625108199979454e-06,
+      "loss": 1.6056,
+      "step": 1835
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.614382131339911e-06,
+      "loss": 1.8108,
+      "step": 1836
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.603662306209671e-06,
+      "loss": 1.7316,
+      "step": 1837
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.592948739849838e-06,
+      "loss": 1.6777,
+      "step": 1838
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.58224144751258e-06,
+      "loss": 1.7195,
+      "step": 1839
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.5715404444411615e-06,
+      "loss": 1.2523,
+      "step": 1840
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.560845745869879e-06,
+      "loss": 1.5651,
+      "step": 1841
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.55015736702405e-06,
+      "loss": 1.7505,
+      "step": 1842
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.539475323120014e-06,
+      "loss": 1.5079,
+      "step": 1843
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.528799629365062e-06,
+      "loss": 1.4314,
+      "step": 1844
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.518130300957476e-06,
+      "loss": 1.6132,
+      "step": 1845
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.507467353086449e-06,
+      "loss": 1.3751,
+      "step": 1846
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.496810800932113e-06,
+      "loss": 1.6673,
+      "step": 1847
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.4861606596654805e-06,
+      "loss": 1.2954,
+      "step": 1848
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.475516944448437e-06,
+      "loss": 1.4659,
+      "step": 1849
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.464879670433738e-06,
+      "loss": 1.3197,
+      "step": 1850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.4542488527649385e-06,
+      "loss": 1.5441,
+      "step": 1851
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.443624506576433e-06,
+      "loss": 1.3866,
+      "step": 1852
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.433006646993379e-06,
+      "loss": 1.5287,
+      "step": 1853
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.422395289131712e-06,
+      "loss": 1.4827,
+      "step": 1854
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.411790448098117e-06,
+      "loss": 1.4118,
+      "step": 1855
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.401192138989979e-06,
+      "loss": 1.399,
+      "step": 1856
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.390600376895413e-06,
+      "loss": 1.5438,
+      "step": 1857
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.380015176893183e-06,
+      "loss": 1.5333,
+      "step": 1858
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.369436554052738e-06,
+      "loss": 1.2251,
+      "step": 1859
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.358864523434148e-06,
+      "loss": 1.6184,
+      "step": 1860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.348299100088101e-06,
+      "loss": 1.3506,
+      "step": 1861
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.3377402990558876e-06,
+      "loss": 1.478,
+      "step": 1862
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.32718813536935e-06,
+      "loss": 1.1999,
+      "step": 1863
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.316642624050905e-06,
+      "loss": 1.5515,
+      "step": 1864
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.306103780113484e-06,
+      "loss": 1.4353,
+      "step": 1865
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.295571618560531e-06,
+      "loss": 1.7013,
+      "step": 1866
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.285046154385976e-06,
+      "loss": 1.6689,
+      "step": 1867
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.274527402574212e-06,
+      "loss": 1.5817,
+      "step": 1868
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.264015378100088e-06,
+      "loss": 1.3292,
+      "step": 1869
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.2535100959288534e-06,
+      "loss": 1.6756,
+      "step": 1870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.243011571016181e-06,
+      "loss": 1.7555,
+      "step": 1871
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.232519818308115e-06,
+      "loss": 1.4063,
+      "step": 1872
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.222034852741054e-06,
+      "loss": 1.3522,
+      "step": 1873
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.211556689241742e-06,
+      "loss": 1.4141,
+      "step": 1874
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.20108534272723e-06,
+      "loss": 1.3918,
+      "step": 1875
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.190620828104878e-06,
+      "loss": 1.4967,
+      "step": 1876
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.180163160272309e-06,
+      "loss": 1.2588,
+      "step": 1877
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1697123541174025e-06,
+      "loss": 1.5597,
+      "step": 1878
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.159268424518264e-06,
+      "loss": 1.2347,
+      "step": 1879
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.14883138634322e-06,
+      "loss": 1.5946,
+      "step": 1880
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.138401254450773e-06,
+      "loss": 1.51,
+      "step": 1881
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.127978043689607e-06,
+      "loss": 1.4591,
+      "step": 1882
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.1175617688985446e-06,
+      "loss": 1.5665,
+      "step": 1883
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.107152444906536e-06,
+      "loss": 1.2853,
+      "step": 1884
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.0967500865326334e-06,
+      "loss": 1.6581,
+      "step": 1885
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.086354708585979e-06,
+      "loss": 1.6065,
+      "step": 1886
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.07596632586577e-06,
+      "loss": 1.4056,
+      "step": 1887
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.0655849531612465e-06,
+      "loss": 1.4406,
+      "step": 1888
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.055210605251678e-06,
+      "loss": 1.2587,
+      "step": 1889
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.0448432969063235e-06,
+      "loss": 1.4093,
+      "step": 1890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.034483042884419e-06,
+      "loss": 1.601,
+      "step": 1891
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.024129857935167e-06,
+      "loss": 1.3906,
+      "step": 1892
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.0137837567976926e-06,
+      "loss": 1.2058,
+      "step": 1893
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.003444754201059e-06,
+      "loss": 1.372,
+      "step": 1894
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.993112864864191e-06,
+      "loss": 1.65,
+      "step": 1895
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.982788103495918e-06,
+      "loss": 1.2757,
+      "step": 1896
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.972470484794906e-06,
+      "loss": 1.3368,
+      "step": 1897
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.962160023449652e-06,
+      "loss": 1.4221,
+      "step": 1898
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.951856734138469e-06,
+      "loss": 1.3768,
+      "step": 1899
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.941560631529452e-06,
+      "loss": 1.1427,
+      "step": 1900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.931271730280482e-06,
+      "loss": 1.3772,
+      "step": 1901
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.92099004503916e-06,
+      "loss": 1.3651,
+      "step": 1902
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.9107155904428435e-06,
+      "loss": 1.3227,
+      "step": 1903
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.900448381118578e-06,
+      "loss": 1.3926,
+      "step": 1904
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.890188431683095e-06,
+      "loss": 1.4526,
+      "step": 1905
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.879935756742805e-06,
+      "loss": 1.2755,
+      "step": 1906
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.869690370893739e-06,
+      "loss": 1.4351,
+      "step": 1907
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.8594522887215776e-06,
+      "loss": 1.5472,
+      "step": 1908
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.849221524801576e-06,
+      "loss": 1.5711,
+      "step": 1909
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.838998093698596e-06,
+      "loss": 1.3841,
+      "step": 1910
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.828782009967044e-06,
+      "loss": 1.2828,
+      "step": 1911
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.81857328815087e-06,
+      "loss": 1.5097,
+      "step": 1912
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.808371942783554e-06,
+      "loss": 1.1843,
+      "step": 1913
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.798177988388052e-06,
+      "loss": 1.3296,
+      "step": 1914
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.7879914394768225e-06,
+      "loss": 1.4215,
+      "step": 1915
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.777812310551767e-06,
+      "loss": 1.4029,
+      "step": 1916
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.767640616104222e-06,
+      "loss": 1.6383,
+      "step": 1917
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.757476370614957e-06,
+      "loss": 1.2297,
+      "step": 1918
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.74731958855411e-06,
+      "loss": 1.2133,
+      "step": 1919
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.737170284381224e-06,
+      "loss": 1.3092,
+      "step": 1920
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.727028472545165e-06,
+      "loss": 1.4945,
+      "step": 1921
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.716894167484161e-06,
+      "loss": 1.5155,
+      "step": 1922
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.706767383625737e-06,
+      "loss": 1.4136,
+      "step": 1923
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6966481353867085e-06,
+      "loss": 1.7102,
+      "step": 1924
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6865364371731825e-06,
+      "loss": 1.2092,
+      "step": 1925
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6764323033804895e-06,
+      "loss": 1.5983,
+      "step": 1926
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.666335748393214e-06,
+      "loss": 1.3997,
+      "step": 1927
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65624678658514e-06,
+      "loss": 1.6628,
+      "step": 1928
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.646165432319242e-06,
+      "loss": 1.5539,
+      "step": 1929
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.636091699947675e-06,
+      "loss": 1.3694,
+      "step": 1930
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.62602560381172e-06,
+      "loss": 1.4263,
+      "step": 1931
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.615967158241814e-06,
+      "loss": 1.5073,
+      "step": 1932
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6059163775574856e-06,
+      "loss": 1.4607,
+      "step": 1933
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.595873276067354e-06,
+      "loss": 1.3914,
+      "step": 1934
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5858378680691085e-06,
+      "loss": 1.6595,
+      "step": 1935
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.575810167849481e-06,
+      "loss": 1.4194,
+      "step": 1936
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.565790189684247e-06,
+      "loss": 1.5239,
+      "step": 1937
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.555777947838158e-06,
+      "loss": 1.4987,
+      "step": 1938
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.545773456564983e-06,
+      "loss": 1.5604,
+      "step": 1939
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.535776730107438e-06,
+      "loss": 1.4635,
+      "step": 1940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.525787782697191e-06,
+      "loss": 1.3534,
+      "step": 1941
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.515806628554835e-06,
+      "loss": 1.3225,
+      "step": 1942
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.505833281889866e-06,
+      "loss": 1.2723,
+      "step": 1943
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.495867756900671e-06,
+      "loss": 1.3468,
+      "step": 1944
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.485910067774498e-06,
+      "loss": 1.0302,
+      "step": 1945
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.475960228687437e-06,
+      "loss": 1.5602,
+      "step": 1946
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.466018253804405e-06,
+      "loss": 1.3989,
+      "step": 1947
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.456084157279125e-06,
+      "loss": 1.2241,
+      "step": 1948
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.446157953254101e-06,
+      "loss": 1.2904,
+      "step": 1949
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.436239655860599e-06,
+      "loss": 1.3973,
+      "step": 1950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.42632927921864e-06,
+      "loss": 1.4488,
+      "step": 1951
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.416426837436956e-06,
+      "loss": 1.5007,
+      "step": 1952
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.406532344612988e-06,
+      "loss": 1.4288,
+      "step": 1953
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.39664581483286e-06,
+      "loss": 1.358,
+      "step": 1954
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.386767262171362e-06,
+      "loss": 1.7036,
+      "step": 1955
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.376896700691919e-06,
+      "loss": 1.2204,
+      "step": 1956
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.367034144446594e-06,
+      "loss": 1.4717,
+      "step": 1957
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.357179607476042e-06,
+      "loss": 1.4448,
+      "step": 1958
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.347333103809504e-06,
+      "loss": 1.296,
+      "step": 1959
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.337494647464785e-06,
+      "loss": 1.3612,
+      "step": 1960
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.327664252448235e-06,
+      "loss": 1.4104,
+      "step": 1961
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.317841932754725e-06,
+      "loss": 1.4945,
+      "step": 1962
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.308027702367628e-06,
+      "loss": 1.4625,
+      "step": 1963
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.298221575258814e-06,
+      "loss": 1.2922,
+      "step": 1964
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.288423565388599e-06,
+      "loss": 1.5656,
+      "step": 1965
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.278633686705753e-06,
+      "loss": 1.5695,
+      "step": 1966
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.268851953147468e-06,
+      "loss": 1.5205,
+      "step": 1967
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.259078378639338e-06,
+      "loss": 1.6002,
+      "step": 1968
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.249312977095352e-06,
+      "loss": 1.5762,
+      "step": 1969
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.239555762417843e-06,
+      "loss": 1.2551,
+      "step": 1970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.229806748497512e-06,
+      "loss": 1.4463,
+      "step": 1971
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2200659492133715e-06,
+      "loss": 1.5712,
+      "step": 1972
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.2103333784327406e-06,
+      "loss": 1.4268,
+      "step": 1973
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.200609050011229e-06,
+      "loss": 1.4475,
+      "step": 1974
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.190892977792704e-06,
+      "loss": 1.4122,
+      "step": 1975
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1811851756092945e-06,
+      "loss": 1.6737,
+      "step": 1976
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.171485657281333e-06,
+      "loss": 1.2418,
+      "step": 1977
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.161794436617381e-06,
+      "loss": 1.6695,
+      "step": 1978
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.152111527414177e-06,
+      "loss": 1.7606,
+      "step": 1979
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.14243694345662e-06,
+      "loss": 1.2592,
+      "step": 1980
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1327706985177775e-06,
+      "loss": 1.2693,
+      "step": 1981
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.123112806358819e-06,
+      "loss": 1.6064,
+      "step": 1982
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.113463280729047e-06,
+      "loss": 1.4412,
+      "step": 1983
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.103822135365837e-06,
+      "loss": 1.4213,
+      "step": 1984
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.094189383994638e-06,
+      "loss": 1.8249,
+      "step": 1985
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.08456504032896e-06,
+      "loss": 1.531,
+      "step": 1986
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.074949118070319e-06,
+      "loss": 1.1268,
+      "step": 1987
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.065341630908273e-06,
+      "loss": 1.2828,
+      "step": 1988
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0557425925203394e-06,
+      "loss": 1.4168,
+      "step": 1989
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.046152016572037e-06,
+      "loss": 1.2655,
+      "step": 1990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.03656991671682e-06,
+      "loss": 1.3362,
+      "step": 1991
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.026996306596073e-06,
+      "loss": 1.5129,
+      "step": 1992
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.017431199839115e-06,
+      "loss": 1.2201,
+      "step": 1993
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.00787461006313e-06,
+      "loss": 1.1742,
+      "step": 1994
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.998326550873203e-06,
+      "loss": 1.5935,
+      "step": 1995
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9887870358622595e-06,
+      "loss": 1.4226,
+      "step": 1996
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.979256078611068e-06,
+      "loss": 1.1554,
+      "step": 1997
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.96973369268821e-06,
+      "loss": 1.4834,
+      "step": 1998
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.960219891650063e-06,
+      "loss": 1.3716,
+      "step": 1999
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.950714689040797e-06,
+      "loss": 1.4048,
+      "step": 2000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.941218098392313e-06,
+      "loss": 1.539,
+      "step": 2001
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.931730133224283e-06,
+      "loss": 1.5156,
+      "step": 2002
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.92225080704408e-06,
+      "loss": 1.4501,
+      "step": 2003
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.912780133346783e-06,
+      "loss": 1.3051,
+      "step": 2004
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.903318125615153e-06,
+      "loss": 1.4459,
+      "step": 2005
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.893864797319612e-06,
+      "loss": 1.4032,
+      "step": 2006
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.884420161918234e-06,
+      "loss": 1.481,
+      "step": 2007
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.874984232856709e-06,
+      "loss": 1.3008,
+      "step": 2008
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.865557023568334e-06,
+      "loss": 1.4493,
+      "step": 2009
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.856138547473993e-06,
+      "loss": 1.5349,
+      "step": 2010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.846728817982137e-06,
+      "loss": 1.3654,
+      "step": 2011
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.837327848488767e-06,
+      "loss": 1.4167,
+      "step": 2012
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.827935652377404e-06,
+      "loss": 1.3823,
+      "step": 2013
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.818552243019098e-06,
+      "loss": 1.3817,
+      "step": 2014
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.8091776337723708e-06,
+      "loss": 1.4042,
+      "step": 2015
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7998118379832236e-06,
+      "loss": 1.7141,
+      "step": 2016
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.790454868985113e-06,
+      "loss": 1.3342,
+      "step": 2017
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7811067400989233e-06,
+      "loss": 1.3006,
+      "step": 2018
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.771767464632956e-06,
+      "loss": 1.562,
+      "step": 2019
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.762437055882916e-06,
+      "loss": 1.4655,
+      "step": 2020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7531155271318744e-06,
+      "loss": 1.4776,
+      "step": 2021
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7438028916502656e-06,
+      "loss": 1.5002,
+      "step": 2022
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.734499162695864e-06,
+      "loss": 1.4091,
+      "step": 2023
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7252043535137606e-06,
+      "loss": 1.3672,
+      "step": 2024
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.715918477336352e-06,
+      "loss": 1.5356,
+      "step": 2025
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.706641547383312e-06,
+      "loss": 1.3502,
+      "step": 2026
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6973735768615894e-06,
+      "loss": 1.1353,
+      "step": 2027
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.688114578965366e-06,
+      "loss": 1.4528,
+      "step": 2028
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.678864566876055e-06,
+      "loss": 1.2521,
+      "step": 2029
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.669623553762278e-06,
+      "loss": 1.4918,
+      "step": 2030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6603915527798382e-06,
+      "loss": 1.5781,
+      "step": 2031
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6511685770717286e-06,
+      "loss": 1.5873,
+      "step": 2032
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6419546397680627e-06,
+      "loss": 1.395,
+      "step": 2033
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6327497539861144e-06,
+      "loss": 1.3087,
+      "step": 2034
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6235539328302584e-06,
+      "loss": 1.6652,
+      "step": 2035
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.614367189391964e-06,
+      "loss": 1.4231,
+      "step": 2036
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.605189536749788e-06,
+      "loss": 1.4437,
+      "step": 2037
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5960209879693263e-06,
+      "loss": 1.2058,
+      "step": 2038
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.586861556103237e-06,
+      "loss": 1.5371,
+      "step": 2039
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.577711254191176e-06,
+      "loss": 1.2755,
+      "step": 2040
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.568570095259821e-06,
+      "loss": 1.2946,
+      "step": 2041
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5594380923228244e-06,
+      "loss": 1.0936,
+      "step": 2042
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5503152583807987e-06,
+      "loss": 1.3753,
+      "step": 2043
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5412016064213216e-06,
+      "loss": 1.3818,
+      "step": 2044
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5320971494188715e-06,
+      "loss": 1.5873,
+      "step": 2045
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5230019003348627e-06,
+      "loss": 1.4284,
+      "step": 2046
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.513915872117586e-06,
+      "loss": 1.2154,
+      "step": 2047
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.504839077702207e-06,
+      "loss": 1.2334,
+      "step": 2048
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.495771530010755e-06,
+      "loss": 1.46,
+      "step": 2049
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.486713241952078e-06,
+      "loss": 1.5319,
+      "step": 2050
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.477664226421862e-06,
+      "loss": 1.1816,
+      "step": 2051
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4686244963025704e-06,
+      "loss": 1.4446,
+      "step": 2052
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4595940644634684e-06,
+      "loss": 1.2212,
+      "step": 2053
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4505729437605727e-06,
+      "loss": 1.4979,
+      "step": 2054
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.441561147036642e-06,
+      "loss": 1.2802,
+      "step": 2055
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4325586871211745e-06,
+      "loss": 1.2958,
+      "step": 2056
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4235655768303542e-06,
+      "loss": 1.1839,
+      "step": 2057
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4145818289670796e-06,
+      "loss": 1.3612,
+      "step": 2058
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.405607456320903e-06,
+      "loss": 1.3424,
+      "step": 2059
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.396642471668037e-06,
+      "loss": 1.3505,
+      "step": 2060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3876868877713253e-06,
+      "loss": 1.3737,
+      "step": 2061
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.378740717380229e-06,
+      "loss": 1.2234,
+      "step": 2062
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3698039732308197e-06,
+      "loss": 1.416,
+      "step": 2063
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.360876668045725e-06,
+      "loss": 1.3574,
+      "step": 2064
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3519588145341586e-06,
+      "loss": 1.3652,
+      "step": 2065
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.343050425391866e-06,
+      "loss": 1.1773,
+      "step": 2066
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3341515133011236e-06,
+      "loss": 1.2722,
+      "step": 2067
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3252620909307123e-06,
+      "loss": 1.3004,
+      "step": 2068
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.316382170935901e-06,
+      "loss": 1.3741,
+      "step": 2069
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3075117659584444e-06,
+      "loss": 1.4394,
+      "step": 2070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2986508886265367e-06,
+      "loss": 1.459,
+      "step": 2071
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.289799551554812e-06,
+      "loss": 1.365,
+      "step": 2072
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.280957767344326e-06,
+      "loss": 1.2966,
+      "step": 2073
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.272125548582531e-06,
+      "loss": 1.2573,
+      "step": 2074
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.263302907843263e-06,
+      "loss": 1.5322,
+      "step": 2075
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2544898576867212e-06,
+      "loss": 1.3625,
+      "step": 2076
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2456864106594564e-06,
+      "loss": 1.4287,
+      "step": 2077
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.236892579294343e-06,
+      "loss": 1.4564,
+      "step": 2078
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2281083761105657e-06,
+      "loss": 1.5942,
+      "step": 2079
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2193338136136055e-06,
+      "loss": 1.7454,
+      "step": 2080
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.210568904295214e-06,
+      "loss": 1.5231,
+      "step": 2081
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.201813660633407e-06,
+      "loss": 1.0115,
+      "step": 2082
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1930680950924375e-06,
+      "loss": 1.3776,
+      "step": 2083
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1843322201227755e-06,
+      "loss": 1.4914,
+      "step": 2084
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1756060481611006e-06,
+      "loss": 1.4918,
+      "step": 2085
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1668895916302765e-06,
+      "loss": 1.46,
+      "step": 2086
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.158182862939334e-06,
+      "loss": 1.3948,
+      "step": 2087
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.149485874483462e-06,
+      "loss": 1.1684,
+      "step": 2088
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.140798638643977e-06,
+      "loss": 1.3323,
+      "step": 2089
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1321211677883122e-06,
+      "loss": 1.6293,
+      "step": 2090
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1234534742699984e-06,
+      "loss": 1.3293,
+      "step": 2091
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1147955704286483e-06,
+      "loss": 1.2917,
+      "step": 2092
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1061474685899386e-06,
+      "loss": 1.5275,
+      "step": 2093
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0975091810655868e-06,
+      "loss": 1.4996,
+      "step": 2094
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.088880720153348e-06,
+      "loss": 1.4054,
+      "step": 2095
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.080262098136979e-06,
+      "loss": 1.2308,
+      "step": 2096
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0716533272862327e-06,
+      "loss": 1.5492,
+      "step": 2097
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.063054419856837e-06,
+      "loss": 1.3727,
+      "step": 2098
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.054465388090475e-06,
+      "loss": 1.4997,
+      "step": 2099
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0458862442147843e-06,
+      "loss": 1.4322,
+      "step": 2100
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.037317000443302e-06,
+      "loss": 1.3417,
+      "step": 2101
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0287576689754926e-06,
+      "loss": 1.2467,
+      "step": 2102
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.0202082619966987e-06,
+      "loss": 1.6833,
+      "step": 2103
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.0116687916781374e-06,
+      "loss": 1.4901,
+      "step": 2104
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.003139270176877e-06,
+      "loss": 1.6554,
+      "step": 2105
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9946197096358197e-06,
+      "loss": 1.2817,
+      "step": 2106
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.986110122183703e-06,
+      "loss": 1.4481,
+      "step": 2107
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.977610519935041e-06,
+      "loss": 1.3326,
+      "step": 2108
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9691209149901543e-06,
+      "loss": 1.3675,
+      "step": 2109
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9606413194351214e-06,
+      "loss": 1.2567,
+      "step": 2110
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9521717453417665e-06,
+      "loss": 1.5993,
+      "step": 2111
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9437122047676648e-06,
+      "loss": 1.5903,
+      "step": 2112
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9352627097560826e-06,
+      "loss": 1.2208,
+      "step": 2113
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.926823272336009e-06,
+      "loss": 1.2143,
+      "step": 2114
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9183939045220923e-06,
+      "loss": 1.3815,
+      "step": 2115
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.909974618314665e-06,
+      "loss": 1.332,
+      "step": 2116
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9015654256996972e-06,
+      "loss": 1.5051,
+      "step": 2117
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8931663386487863e-06,
+      "loss": 1.1522,
+      "step": 2118
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8847773691191596e-06,
+      "loss": 1.3759,
+      "step": 2119
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8763985290536135e-06,
+      "loss": 1.4331,
+      "step": 2120
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.868029830380551e-06,
+      "loss": 1.4589,
+      "step": 2121
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.859671285013922e-06,
+      "loss": 1.4327,
+      "step": 2122
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.851322904853222e-06,
+      "loss": 1.323,
+      "step": 2123
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8429847017834877e-06,
+      "loss": 1.4878,
+      "step": 2124
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.834656687675247e-06,
+      "loss": 1.5697,
+      "step": 2125
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.826338874384541e-06,
+      "loss": 1.4001,
+      "step": 2126
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8180312737528804e-06,
+      "loss": 1.0184,
+      "step": 2127
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8097338976072362e-06,
+      "loss": 1.3859,
+      "step": 2128
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.801446757760026e-06,
+      "loss": 1.3821,
+      "step": 2129
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.79316986600909e-06,
+      "loss": 1.5042,
+      "step": 2130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7849032341376914e-06,
+      "loss": 1.4607,
+      "step": 2131
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.776646873914466e-06,
+      "loss": 1.5038,
+      "step": 2132
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.768400797093449e-06,
+      "loss": 1.2624,
+      "step": 2133
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.76016501541402e-06,
+      "loss": 1.3989,
+      "step": 2134
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7519395406009107e-06,
+      "loss": 1.493,
+      "step": 2135
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.743724384364176e-06,
+      "loss": 1.4691,
+      "step": 2136
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.735519558399177e-06,
+      "loss": 1.7036,
+      "step": 2137
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.727325074386579e-06,
+      "loss": 1.3342,
+      "step": 2138
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.719140943992317e-06,
+      "loss": 1.4529,
+      "step": 2139
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.7109671788675838e-06,
+      "loss": 1.3309,
+      "step": 2140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.7028037906488223e-06,
+      "loss": 1.227,
+      "step": 2141
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.694650790957698e-06,
+      "loss": 1.4928,
+      "step": 2142
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6865081914010894e-06,
+      "loss": 1.2924,
+      "step": 2143
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6783760035710636e-06,
+      "loss": 1.4359,
+      "step": 2144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6702542390448773e-06,
+      "loss": 1.3516,
+      "step": 2145
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.662142909384937e-06,
+      "loss": 1.3663,
+      "step": 2146
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6540420261387966e-06,
+      "loss": 1.6022,
+      "step": 2147
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6459516008391394e-06,
+      "loss": 1.6489,
+      "step": 2148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.63787164500376e-06,
+      "loss": 1.3224,
+      "step": 2149
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.629802170135546e-06,
+      "loss": 1.472,
+      "step": 2150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.621743187722472e-06,
+      "loss": 1.5771,
+      "step": 2151
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6136947092375655e-06,
+      "loss": 1.1191,
+      "step": 2152
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.605656746138905e-06,
+      "loss": 1.3474,
+      "step": 2153
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.597629309869597e-06,
+      "loss": 1.3218,
+      "step": 2154
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.589612411857765e-06,
+      "loss": 1.6099,
+      "step": 2155
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5816060635165254e-06,
+      "loss": 1.5729,
+      "step": 2156
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5736102762439775e-06,
+      "loss": 1.3291,
+      "step": 2157
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5656250614231893e-06,
+      "loss": 1.2772,
+      "step": 2158
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5576504304221727e-06,
+      "loss": 1.4413,
+      "step": 2159
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.549686394593874e-06,
+      "loss": 1.3817,
+      "step": 2160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.541732965276156e-06,
+      "loss": 1.3927,
+      "step": 2161
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5337901537917776e-06,
+      "loss": 1.3298,
+      "step": 2162
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5258579714483933e-06,
+      "loss": 1.4091,
+      "step": 2163
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.517936429538508e-06,
+      "loss": 1.4948,
+      "step": 2164
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5100255393394956e-06,
+      "loss": 1.2598,
+      "step": 2165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.502125312113556e-06,
+      "loss": 1.4503,
+      "step": 2166
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4942357591077125e-06,
+      "loss": 1.2476,
+      "step": 2167
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4863568915537893e-06,
+      "loss": 1.4236,
+      "step": 2168
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4784887206684007e-06,
+      "loss": 1.6978,
+      "step": 2169
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.470631257652939e-06,
+      "loss": 1.3295,
+      "step": 2170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4627845136935336e-06,
+      "loss": 1.4273,
+      "step": 2171
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4549484999610773e-06,
+      "loss": 1.3365,
+      "step": 2172
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4471232276111723e-06,
+      "loss": 1.143,
+      "step": 2173
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4393087077841293e-06,
+      "loss": 1.5348,
+      "step": 2174
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.431504951604964e-06,
+      "loss": 1.6333,
+      "step": 2175
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.423711970183347e-06,
+      "loss": 1.3676,
+      "step": 2176
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.415929774613631e-06,
+      "loss": 1.2428,
+      "step": 2177
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.408158375974804e-06,
+      "loss": 1.3881,
+      "step": 2178
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.4003977853304774e-06,
+      "loss": 1.3635,
+      "step": 2179
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.392648013728893e-06,
+      "loss": 1.4889,
+      "step": 2180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.384909072202868e-06,
+      "loss": 1.5048,
+      "step": 2181
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3771809717698233e-06,
+      "loss": 1.4755,
+      "step": 2182
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3694637234317252e-06,
+      "loss": 1.1849,
+      "step": 2183
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3617573381751114e-06,
+      "loss": 1.5121,
+      "step": 2184
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3540618269710393e-06,
+      "loss": 1.3481,
+      "step": 2185
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3463772007750898e-06,
+      "loss": 1.313,
+      "step": 2186
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.338703470527355e-06,
+      "loss": 1.3664,
+      "step": 2187
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3310406471523994e-06,
+      "loss": 1.2965,
+      "step": 2188
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.323388741559277e-06,
+      "loss": 1.5522,
+      "step": 2189
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3157477646414896e-06,
+      "loss": 1.3192,
+      "step": 2190
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3081177272769806e-06,
+      "loss": 1.2854,
+      "step": 2191
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3004986403281215e-06,
+      "loss": 1.4903,
+      "step": 2192
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2928905146416925e-06,
+      "loss": 1.422,
+      "step": 2193
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.285293361048878e-06,
+      "loss": 1.3284,
+      "step": 2194
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2777071903652248e-06,
+      "loss": 1.5621,
+      "step": 2195
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2701320133906614e-06,
+      "loss": 1.5547,
+      "step": 2196
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2625678409094554e-06,
+      "loss": 1.424,
+      "step": 2197
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2550146836902107e-06,
+      "loss": 1.5292,
+      "step": 2198
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.247472552485849e-06,
+      "loss": 1.5616,
+      "step": 2199
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.239941458033593e-06,
+      "loss": 1.3659,
+      "step": 2200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2324214110549613e-06,
+      "loss": 1.6546,
+      "step": 2201
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2249124222557362e-06,
+      "loss": 1.4161,
+      "step": 2202
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2174145023259595e-06,
+      "loss": 1.4106,
+      "step": 2203
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.209927661939918e-06,
+      "loss": 1.258,
+      "step": 2204
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2024519117561203e-06,
+      "loss": 1.5043,
+      "step": 2205
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1949872624172907e-06,
+      "loss": 1.388,
+      "step": 2206
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.187533724550346e-06,
+      "loss": 1.4317,
+      "step": 2207
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.180091308766391e-06,
+      "loss": 1.29,
+      "step": 2208
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.172660025660692e-06,
+      "loss": 1.3285,
+      "step": 2209
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1652398858126656e-06,
+      "loss": 1.6029,
+      "step": 2210
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1578308997858664e-06,
+      "loss": 1.3068,
+      "step": 2211
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1504330781279703e-06,
+      "loss": 1.6577,
+      "step": 2212
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1430464313707557e-06,
+      "loss": 1.4977,
+      "step": 2213
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1356709700300994e-06,
+      "loss": 1.2976,
+      "step": 2214
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.128306704605949e-06,
+      "loss": 1.2786,
+      "step": 2215
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1209536455823123e-06,
+      "loss": 1.3484,
+      "step": 2216
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1136118034272444e-06,
+      "loss": 1.1646,
+      "step": 2217
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.106281188592836e-06,
+      "loss": 1.5719,
+      "step": 2218
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.098961811515187e-06,
+      "loss": 1.5823,
+      "step": 2219
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0916536826144006e-06,
+      "loss": 1.3301,
+      "step": 2220
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0843568122945747e-06,
+      "loss": 1.5403,
+      "step": 2221
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.077071210943771e-06,
+      "loss": 1.4609,
+      "step": 2222
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.069796888934008e-06,
+      "loss": 1.1364,
+      "step": 2223
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0625338566212505e-06,
+      "loss": 1.2703,
+      "step": 2224
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.055282124345387e-06,
+      "loss": 1.7914,
+      "step": 2225
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.048041702430228e-06,
+      "loss": 1.0629,
+      "step": 2226
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.040812601183465e-06,
+      "loss": 1.3095,
+      "step": 2227
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.033594830896691e-06,
+      "loss": 1.2865,
+      "step": 2228
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0263884018453596e-06,
+      "loss": 1.6661,
+      "step": 2229
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.019193324288775e-06,
+      "loss": 1.4554,
+      "step": 2230
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0120096084700936e-06,
+      "loss": 1.6451,
+      "step": 2231
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.0048372646162774e-06,
+      "loss": 1.3967,
+      "step": 2232
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9976763029381197e-06,
+      "loss": 1.4399,
+      "step": 2233
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.990526733630196e-06,
+      "loss": 1.5767,
+      "step": 2234
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9833885668708686e-06,
+      "loss": 1.1332,
+      "step": 2235
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9762618128222653e-06,
+      "loss": 1.1133,
+      "step": 2236
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.969146481630263e-06,
+      "loss": 1.4982,
+      "step": 2237
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9620425834244893e-06,
+      "loss": 1.6646,
+      "step": 2238
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.954950128318276e-06,
+      "loss": 1.4872,
+      "step": 2239
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9478691264086824e-06,
+      "loss": 1.5388,
+      "step": 2240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.940799587776452e-06,
+      "loss": 1.3563,
+      "step": 2241
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.933741522486009e-06,
+      "loss": 1.2739,
+      "step": 2242
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.926694940585454e-06,
+      "loss": 1.2479,
+      "step": 2243
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9196598521065212e-06,
+      "loss": 1.4168,
+      "step": 2244
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.912636267064605e-06,
+      "loss": 1.1941,
+      "step": 2245
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9056241954586997e-06,
+      "loss": 1.5037,
+      "step": 2246
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8986236472714281e-06,
+      "loss": 1.5659,
+      "step": 2247
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8916346324689982e-06,
+      "loss": 1.2473,
+      "step": 2248
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8846571610011965e-06,
+      "loss": 1.3946,
+      "step": 2249
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8776912428013883e-06,
+      "loss": 1.5602,
+      "step": 2250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8707368877864729e-06,
+      "loss": 1.3332,
+      "step": 2251
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8637941058569065e-06,
+      "loss": 1.6887,
+      "step": 2252
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.856862906896657e-06,
+      "loss": 1.4142,
+      "step": 2253
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8499433007732037e-06,
+      "loss": 1.2743,
+      "step": 2254
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8430352973375344e-06,
+      "loss": 1.2524,
+      "step": 2255
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8361389064240965e-06,
+      "loss": 1.4024,
+      "step": 2256
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.829254137850829e-06,
+      "loss": 1.3586,
+      "step": 2257
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8223810014191046e-06,
+      "loss": 1.6931,
+      "step": 2258
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.815519506913752e-06,
+      "loss": 1.3345,
+      "step": 2259
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.808669664103019e-06,
+      "loss": 1.5224,
+      "step": 2260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8018314827385618e-06,
+      "loss": 1.4597,
+      "step": 2261
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7950049725554476e-06,
+      "loss": 1.4612,
+      "step": 2262
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7881901432721104e-06,
+      "loss": 1.3099,
+      "step": 2263
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7813870045903715e-06,
+      "loss": 1.4219,
+      "step": 2264
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7745955661953996e-06,
+      "loss": 1.3978,
+      "step": 2265
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7678158377557097e-06,
+      "loss": 1.4722,
+      "step": 2266
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7610478289231435e-06,
+      "loss": 1.4937,
+      "step": 2267
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7542915493328594e-06,
+      "loss": 1.4675,
+      "step": 2268
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7475470086033264e-06,
+      "loss": 1.5819,
+      "step": 2269
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7408142163362818e-06,
+      "loss": 1.4146,
+      "step": 2270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7340931821167572e-06,
+      "loss": 1.3415,
+      "step": 2271
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7273839155130355e-06,
+      "loss": 1.5336,
+      "step": 2272
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7206864260766486e-06,
+      "loss": 1.4796,
+      "step": 2273
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7140007233423606e-06,
+      "loss": 1.5048,
+      "step": 2274
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7073268168281564e-06,
+      "loss": 1.5898,
+      "step": 2275
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7006647160352308e-06,
+      "loss": 1.4885,
+      "step": 2276
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6940144304479678e-06,
+      "loss": 1.2994,
+      "step": 2277
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.687375969533932e-06,
+      "loss": 1.4332,
+      "step": 2278
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6807493427438526e-06,
+      "loss": 1.3831,
+      "step": 2279
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6741345595116133e-06,
+      "loss": 1.4273,
+      "step": 2280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6675316292542344e-06,
+      "loss": 1.0097,
+      "step": 2281
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6609405613718688e-06,
+      "loss": 1.3448,
+      "step": 2282
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.654361365247773e-06,
+      "loss": 1.5362,
+      "step": 2283
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.647794050248307e-06,
+      "loss": 1.3242,
+      "step": 2284
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6412386257229152e-06,
+      "loss": 1.3754,
+      "step": 2285
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6346951010041146e-06,
+      "loss": 1.7743,
+      "step": 2286
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6281634854074823e-06,
+      "loss": 1.568,
+      "step": 2287
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6216437882316382e-06,
+      "loss": 1.2589,
+      "step": 2288
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.61513601875824e-06,
+      "loss": 1.5652,
+      "step": 2289
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6086401862519608e-06,
+      "loss": 1.793,
+      "step": 2290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6021562999604789e-06,
+      "loss": 1.4683,
+      "step": 2291
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5956843691144686e-06,
+      "loss": 1.6087,
+      "step": 2292
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5892244029275805e-06,
+      "loss": 1.3121,
+      "step": 2293
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5827764105964405e-06,
+      "loss": 1.2332,
+      "step": 2294
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5763404013006124e-06,
+      "loss": 1.3393,
+      "step": 2295
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5699163842026166e-06,
+      "loss": 1.4695,
+      "step": 2296
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5635043684478933e-06,
+      "loss": 1.4767,
+      "step": 2297
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5571043631647976e-06,
+      "loss": 1.3673,
+      "step": 2298
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5507163774645862e-06,
+      "loss": 1.454,
+      "step": 2299
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5443404204414025e-06,
+      "loss": 1.3161,
+      "step": 2300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5379765011722758e-06,
+      "loss": 1.6034,
+      "step": 2301
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.531624628717081e-06,
+      "loss": 1.4068,
+      "step": 2302
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.525284812118557e-06,
+      "loss": 1.3525,
+      "step": 2303
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.518957060402274e-06,
+      "loss": 1.4532,
+      "step": 2304
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.512641382576624e-06,
+      "loss": 1.4955,
+      "step": 2305
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.506337787632819e-06,
+      "loss": 1.3729,
+      "step": 2306
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5000462845448537e-06,
+      "loss": 1.7372,
+      "step": 2307
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4937668822695294e-06,
+      "loss": 1.6803,
+      "step": 2308
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4874995897463972e-06,
+      "loss": 1.2643,
+      "step": 2309
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.481244415897789e-06,
+      "loss": 1.6122,
+      "step": 2310
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4750013696287723e-06,
+      "loss": 1.4969,
+      "step": 2311
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4687704598271502e-06,
+      "loss": 1.4946,
+      "step": 2312
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4625516953634567e-06,
+      "loss": 1.4947,
+      "step": 2313
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.456345085090919e-06,
+      "loss": 1.5622,
+      "step": 2314
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4501506378454787e-06,
+      "loss": 1.0848,
+      "step": 2315
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.44396836244575e-06,
+      "loss": 1.3604,
+      "step": 2316
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4377982676930235e-06,
+      "loss": 1.521,
+      "step": 2317
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4316403623712516e-06,
+      "loss": 1.5201,
+      "step": 2318
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.425494655247024e-06,
+      "loss": 1.3448,
+      "step": 2319
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4193611550695773e-06,
+      "loss": 1.4419,
+      "step": 2320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.413239870570756e-06,
+      "loss": 1.6954,
+      "step": 2321
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4071308104650272e-06,
+      "loss": 1.5843,
+      "step": 2322
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4010339834494468e-06,
+      "loss": 1.7759,
+      "step": 2323
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3949493982036555e-06,
+      "loss": 1.3358,
+      "step": 2324
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3888770633898762e-06,
+      "loss": 1.3894,
+      "step": 2325
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.382816987652873e-06,
+      "loss": 1.4051,
+      "step": 2326
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3767691796199745e-06,
+      "loss": 1.7752,
+      "step": 2327
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3707336479010381e-06,
+      "loss": 1.4798,
+      "step": 2328
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3647104010884438e-06,
+      "loss": 1.6275,
+      "step": 2329
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3586994477570825e-06,
+      "loss": 1.4139,
+      "step": 2330
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3527007964643436e-06,
+      "loss": 1.5555,
+      "step": 2331
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3467144557501065e-06,
+      "loss": 1.7837,
+      "step": 2332
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3407404341367203e-06,
+      "loss": 1.4145,
+      "step": 2333
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.334778740128998e-06,
+      "loss": 1.4779,
+      "step": 2334
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3288293822142017e-06,
+      "loss": 1.4364,
+      "step": 2335
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3228923688620342e-06,
+      "loss": 1.4961,
+      "step": 2336
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3169677085246213e-06,
+      "loss": 1.514,
+      "step": 2337
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3110554096365003e-06,
+      "loss": 1.5074,
+      "step": 2338
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3051554806146195e-06,
+      "loss": 1.4463,
+      "step": 2339
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2992679298583089e-06,
+      "loss": 1.6143,
+      "step": 2340
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2933927657492774e-06,
+      "loss": 1.3421,
+      "step": 2341
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.287529996651602e-06,
+      "loss": 1.5018,
+      "step": 2342
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.281679630911714e-06,
+      "loss": 1.2337,
+      "step": 2343
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2758416768583814e-06,
+      "loss": 1.3933,
+      "step": 2344
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2700161428027124e-06,
+      "loss": 1.1947,
+      "step": 2345
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2642030370381264e-06,
+      "loss": 1.4681,
+      "step": 2346
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2584023678403502e-06,
+      "loss": 1.7516,
+      "step": 2347
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2526141434674076e-06,
+      "loss": 1.3862,
+      "step": 2348
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2468383721596044e-06,
+      "loss": 1.7484,
+      "step": 2349
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2410750621395163e-06,
+      "loss": 1.4472,
+      "step": 2350
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2353242216119798e-06,
+      "loss": 1.3695,
+      "step": 2351
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.229585858764084e-06,
+      "loss": 1.4722,
+      "step": 2352
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2238599817651486e-06,
+      "loss": 1.3181,
+      "step": 2353
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2181465987667174e-06,
+      "loss": 1.4107,
+      "step": 2354
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2124457179025527e-06,
+      "loss": 1.2216,
+      "step": 2355
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2067573472886108e-06,
+      "loss": 1.526,
+      "step": 2356
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2010814950230498e-06,
+      "loss": 1.392,
+      "step": 2357
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.195418169186191e-06,
+      "loss": 1.3603,
+      "step": 2358
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1897673778405372e-06,
+      "loss": 1.3015,
+      "step": 2359
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1841291290307356e-06,
+      "loss": 1.355,
+      "step": 2360
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1785034307835853e-06,
+      "loss": 1.5695,
+      "step": 2361
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.172890291108012e-06,
+      "loss": 1.3653,
+      "step": 2362
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1672897179950648e-06,
+      "loss": 1.6426,
+      "step": 2363
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.161701719417908e-06,
+      "loss": 1.3159,
+      "step": 2364
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1561263033317926e-06,
+      "loss": 1.5476,
+      "step": 2365
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1505634776740693e-06,
+      "loss": 1.3,
+      "step": 2366
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1450132503641564e-06,
+      "loss": 1.3006,
+      "step": 2367
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1394756293035369e-06,
+      "loss": 1.3904,
+      "step": 2368
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1339506223757579e-06,
+      "loss": 1.2846,
+      "step": 2369
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.128438237446391e-06,
+      "loss": 1.5087,
+      "step": 2370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1229384823630552e-06,
+      "loss": 1.4214,
+      "step": 2371
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1174513649553742e-06,
+      "loss": 1.2962,
+      "step": 2372
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.111976893034996e-06,
+      "loss": 1.2474,
+      "step": 2373
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1065150743955544e-06,
+      "loss": 1.4587,
+      "step": 2374
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1010659168126713e-06,
+      "loss": 1.4724,
+      "step": 2375
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0956294280439527e-06,
+      "loss": 1.3681,
+      "step": 2376
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0902056158289542e-06,
+      "loss": 1.5189,
+      "step": 2377
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.084794487889199e-06,
+      "loss": 1.7258,
+      "step": 2378
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0793960519281433e-06,
+      "loss": 1.4144,
+      "step": 2379
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0740103156311753e-06,
+      "loss": 1.2554,
+      "step": 2380
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0686372866656124e-06,
+      "loss": 1.665,
+      "step": 2381
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.063276972680667e-06,
+      "loss": 1.2705,
+      "step": 2382
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.057929381307462e-06,
+      "loss": 1.382,
+      "step": 2383
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0525945201590037e-06,
+      "loss": 1.3876,
+      "step": 2384
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0472723968301712e-06,
+      "loss": 1.1645,
+      "step": 2385
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0419630188977213e-06,
+      "loss": 1.5499,
+      "step": 2386
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0366663939202515e-06,
+      "loss": 1.3362,
+      "step": 2387
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.031382529438215e-06,
+      "loss": 1.5812,
+      "step": 2388
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0261114329738897e-06,
+      "loss": 1.2592,
+      "step": 2389
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0208531120313857e-06,
+      "loss": 1.5211,
+      "step": 2390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0156075740966198e-06,
+      "loss": 1.2363,
+      "step": 2391
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.010374826637308e-06,
+      "loss": 1.6139,
+      "step": 2392
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0051548771029696e-06,
+      "loss": 1.3463,
+      "step": 2393
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.999477329248864e-07,
+      "loss": 1.3339,
+      "step": 2394
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.947534015161254e-07,
+      "loss": 1.374,
+      "step": 2395
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.895718902715057e-07,
+      "loss": 1.0881,
+      "step": 2396
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.844032065675945e-07,
+      "loss": 1.0885,
+      "step": 2397
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.792473577627026e-07,
+      "loss": 1.2909,
+      "step": 2398
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.741043511968605e-07,
+      "loss": 1.6412,
+      "step": 2399
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.689741941918273e-07,
+      "loss": 1.3653,
+      "step": 2400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.638568940510563e-07,
+      "loss": 1.2859,
+      "step": 2401
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.587524580597086e-07,
+      "loss": 1.4281,
+      "step": 2402
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.536608934846236e-07,
+      "loss": 1.2901,
+      "step": 2403
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.4858220757432e-07,
+      "loss": 1.6859,
+      "step": 2404
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.435164075589832e-07,
+      "loss": 1.6743,
+      "step": 2405
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.384635006504483e-07,
+      "loss": 1.2505,
+      "step": 2406
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.334234940422027e-07,
+      "loss": 1.3083,
+      "step": 2407
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.283963949093655e-07,
+      "loss": 1.3136,
+      "step": 2408
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.233822104086765e-07,
+      "loss": 1.4036,
+      "step": 2409
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.183809476784955e-07,
+      "loss": 1.5002,
+      "step": 2410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.133926138387827e-07,
+      "loss": 1.5041,
+      "step": 2411
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.084172159910942e-07,
+      "loss": 1.3798,
+      "step": 2412
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.034547612185673e-07,
+      "loss": 1.3899,
+      "step": 2413
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.985052565859198e-07,
+      "loss": 1.549,
+      "step": 2414
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.935687091394251e-07,
+      "loss": 1.3845,
+      "step": 2415
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.886451259069151e-07,
+      "loss": 1.5776,
+      "step": 2416
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.837345138977638e-07,
+      "loss": 1.6024,
+      "step": 2417
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.788368801028801e-07,
+      "loss": 1.3396,
+      "step": 2418
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.739522314946936e-07,
+      "loss": 1.3144,
+      "step": 2419
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.690805750271536e-07,
+      "loss": 1.3015,
+      "step": 2420
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.642219176357081e-07,
+      "loss": 1.369,
+      "step": 2421
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.593762662373018e-07,
+      "loss": 1.195,
+      "step": 2422
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.545436277303609e-07,
+      "loss": 1.522,
+      "step": 2423
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.497240089947901e-07,
+      "loss": 1.5346,
+      "step": 2424
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.449174168919549e-07,
+      "loss": 1.2527,
+      "step": 2425
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.401238582646775e-07,
+      "loss": 1.21,
+      "step": 2426
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.353433399372257e-07,
+      "loss": 1.5814,
+      "step": 2427
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.305758687153032e-07,
+      "loss": 1.2182,
+      "step": 2428
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.258214513860363e-07,
+      "loss": 1.2452,
+      "step": 2429
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.210800947179698e-07,
+      "loss": 1.3678,
+      "step": 2430
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.163518054610531e-07,
+      "loss": 1.4426,
+      "step": 2431
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.116365903466394e-07,
+      "loss": 1.469,
+      "step": 2432
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.069344560874548e-07,
+      "loss": 1.431,
+      "step": 2433
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.022454093776178e-07,
+      "loss": 1.5834,
+      "step": 2434
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.975694568926085e-07,
+      "loss": 1.5362,
+      "step": 2435
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.92906605289262e-07,
+      "loss": 1.5635,
+      "step": 2436
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.882568612057728e-07,
+      "loss": 1.1414,
+      "step": 2437
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.836202312616626e-07,
+      "loss": 1.8295,
+      "step": 2438
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.789967220577965e-07,
+      "loss": 1.2892,
+      "step": 2439
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.743863401763463e-07,
+      "loss": 1.3894,
+      "step": 2440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.697890921808082e-07,
+      "loss": 1.4946,
+      "step": 2441
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.652049846159726e-07,
+      "loss": 1.5447,
+      "step": 2442
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.606340240079257e-07,
+      "loss": 1.265,
+      "step": 2443
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.56076216864039e-07,
+      "loss": 1.3459,
+      "step": 2444
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.515315696729519e-07,
+      "loss": 1.4567,
+      "step": 2445
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.470000889045758e-07,
+      "loss": 1.2605,
+      "step": 2446
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.424817810100749e-07,
+      "loss": 1.4518,
+      "step": 2447
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.37976652421859e-07,
+      "loss": 1.2771,
+      "step": 2448
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.334847095535813e-07,
+      "loss": 1.2964,
+      "step": 2449
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.290059588001119e-07,
+      "loss": 1.8313,
+      "step": 2450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.24540406537555e-07,
+      "loss": 1.4013,
+      "step": 2451
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.200880591232084e-07,
+      "loss": 1.4062,
+      "step": 2452
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.156489228955866e-07,
+      "loss": 1.569,
+      "step": 2453
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.112230041743862e-07,
+      "loss": 1.6697,
+      "step": 2454
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.068103092604894e-07,
+      "loss": 1.5741,
+      "step": 2455
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.02410844435959e-07,
+      "loss": 1.3958,
+      "step": 2456
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.980246159640092e-07,
+      "loss": 1.4013,
+      "step": 2457
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.936516300890239e-07,
+      "loss": 1.4053,
+      "step": 2458
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.892918930365289e-07,
+      "loss": 1.4796,
+      "step": 2459
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.84945411013186e-07,
+      "loss": 1.6008,
+      "step": 2460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.806121902067919e-07,
+      "loss": 1.4403,
+      "step": 2461
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.76292236786259e-07,
+      "loss": 1.3376,
+      "step": 2462
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.719855569016187e-07,
+      "loss": 1.4014,
+      "step": 2463
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.676921566839977e-07,
+      "loss": 1.2418,
+      "step": 2464
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.634120422456225e-07,
+      "loss": 1.3975,
+      "step": 2465
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.591452196798053e-07,
+      "loss": 1.5146,
+      "step": 2466
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.548916950609341e-07,
+      "loss": 1.2588,
+      "step": 2467
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.506514744444658e-07,
+      "loss": 1.678,
+      "step": 2468
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.464245638669154e-07,
+      "loss": 1.5591,
+      "step": 2469
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.422109693458545e-07,
+      "loss": 1.4771,
+      "step": 2470
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.380106968798927e-07,
+      "loss": 1.544,
+      "step": 2471
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.338237524486756e-07,
+      "loss": 1.5229,
+      "step": 2472
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.296501420128753e-07,
+      "loss": 1.2298,
+      "step": 2473
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.254898715141788e-07,
+      "loss": 1.3072,
+      "step": 2474
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.21342946875283e-07,
+      "loss": 1.2121,
+      "step": 2475
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.172093739998897e-07,
+      "loss": 1.452,
+      "step": 2476
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.13089158772685e-07,
+      "loss": 1.2103,
+      "step": 2477
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.089823070593437e-07,
+      "loss": 1.5263,
+      "step": 2478
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.048888247065143e-07,
+      "loss": 1.3849,
+      "step": 2479
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.008087175418131e-07,
+      "loss": 1.3651,
+      "step": 2480
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.967419913738126e-07,
+      "loss": 1.1303,
+      "step": 2481
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.926886519920372e-07,
+      "loss": 1.4894,
+      "step": 2482
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.886487051669554e-07,
+      "loss": 1.5404,
+      "step": 2483
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.846221566499688e-07,
+      "loss": 1.2261,
+      "step": 2484
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.80609012173401e-07,
+      "loss": 1.1054,
+      "step": 2485
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.766092774504983e-07,
+      "loss": 1.3206,
+      "step": 2486
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.726229581754117e-07,
+      "loss": 1.7983,
+      "step": 2487
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.686500600232003e-07,
+      "loss": 1.523,
+      "step": 2488
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.646905886498055e-07,
+      "loss": 1.2592,
+      "step": 2489
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.607445496920661e-07,
+      "loss": 1.637,
+      "step": 2490
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.568119487676904e-07,
+      "loss": 1.462,
+      "step": 2491
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.528927914752579e-07,
+      "loss": 1.4331,
+      "step": 2492
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.489870833942102e-07,
+      "loss": 1.3316,
+      "step": 2493
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.450948300848379e-07,
+      "loss": 1.522,
+      "step": 2494
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.412160370882868e-07,
+      "loss": 1.5579,
+      "step": 2495
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.373507099265274e-07,
+      "loss": 1.5086,
+      "step": 2496
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.3349885410237e-07,
+      "loss": 1.3342,
+      "step": 2497
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.296604750994416e-07,
+      "loss": 1.2813,
+      "step": 2498
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.258355783821822e-07,
+      "loss": 1.0772,
+      "step": 2499
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.22024169395845e-07,
+      "loss": 1.2934,
+      "step": 2500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.182262535664706e-07,
+      "loss": 1.3594,
+      "step": 2501
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.14441836300903e-07,
+      "loss": 1.4169,
+      "step": 2502
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.106709229867568e-07,
+      "loss": 1.397,
+      "step": 2503
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.069135189924312e-07,
+      "loss": 1.371,
+      "step": 2504
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.031696296670885e-07,
+      "loss": 1.073,
+      "step": 2505
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.994392603406507e-07,
+      "loss": 1.3909,
+      "step": 2506
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.957224163237995e-07,
+      "loss": 1.5981,
+      "step": 2507
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.92019102907948e-07,
+      "loss": 1.2035,
+      "step": 2508
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.883293253652599e-07,
+      "loss": 1.3625,
+      "step": 2509
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.846530889486222e-07,
+      "loss": 1.4266,
+      "step": 2510
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.809903988916431e-07,
+      "loss": 1.4738,
+      "step": 2511
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.773412604086536e-07,
+      "loss": 1.2482,
+      "step": 2512
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.737056786946803e-07,
+      "loss": 1.403,
+      "step": 2513
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7008365892546314e-07,
+      "loss": 1.1934,
+      "step": 2514
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6647520625742184e-07,
+      "loss": 1.4173,
+      "step": 2515
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.628803258276715e-07,
+      "loss": 1.4158,
+      "step": 2516
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.592990227540006e-07,
+      "loss": 1.2161,
+      "step": 2517
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.557313021348697e-07,
+      "loss": 1.5801,
+      "step": 2518
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.521771690494048e-07,
+      "loss": 1.6454,
+      "step": 2519
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.486366285573818e-07,
+      "loss": 1.2621,
+      "step": 2520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.451096856992343e-07,
+      "loss": 1.5782,
+      "step": 2521
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4159634549603145e-07,
+      "loss": 1.3241,
+      "step": 2522
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3809661294948124e-07,
+      "loss": 1.5076,
+      "step": 2523
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3461049304191483e-07,
+      "loss": 1.4416,
+      "step": 2524
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3113799073628894e-07,
+      "loss": 1.2328,
+      "step": 2525
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.276791109761713e-07,
+      "loss": 1.3776,
+      "step": 2526
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2423385868573643e-07,
+      "loss": 1.5137,
+      "step": 2527
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.208022387697586e-07,
+      "loss": 1.4593,
+      "step": 2528
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1738425611360435e-07,
+      "loss": 1.515,
+      "step": 2529
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.139799155832247e-07,
+      "loss": 1.758,
+      "step": 2530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.105892220251517e-07,
+      "loss": 1.6762,
+      "step": 2531
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0721218026648633e-07,
+      "loss": 1.5668,
+      "step": 2532
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.038487951148973e-07,
+      "loss": 1.3416,
+      "step": 2533
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0049907135860986e-07,
+      "loss": 1.146,
+      "step": 2534
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9716301376639955e-07,
+      "loss": 1.3017,
+      "step": 2535
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.938406270875883e-07,
+      "loss": 1.5709,
+      "step": 2536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.905319160520349e-07,
+      "loss": 1.3528,
+      "step": 2537
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.872368853701258e-07,
+      "loss": 1.6733,
+      "step": 2538
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.8395553973277876e-07,
+      "loss": 1.1098,
+      "step": 2539
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.806878838114225e-07,
+      "loss": 1.2154,
+      "step": 2540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7743392225800036e-07,
+      "loss": 1.3542,
+      "step": 2541
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.741936597049578e-07,
+      "loss": 1.4238,
+      "step": 2542
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.709671007652393e-07,
+      "loss": 1.2855,
+      "step": 2543
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6775425003227725e-07,
+      "loss": 1.4152,
+      "step": 2544
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6455511207999504e-07,
+      "loss": 1.1686,
+      "step": 2545
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6136969146278953e-07,
+      "loss": 1.4802,
+      "step": 2546
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.581979927155288e-07,
+      "loss": 1.4657,
+      "step": 2547
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.550400203535476e-07,
+      "loss": 1.3599,
+      "step": 2548
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.518957788726374e-07,
+      "loss": 1.348,
+      "step": 2549
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.487652727490454e-07,
+      "loss": 1.685,
+      "step": 2550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4564850643946214e-07,
+      "loss": 1.4447,
+      "step": 2551
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.425454843810183e-07,
+      "loss": 1.5913,
+      "step": 2552
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.394562109912769e-07,
+      "loss": 1.3624,
+      "step": 2553
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.36380690668231e-07,
+      "loss": 1.2554,
+      "step": 2554
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.333189277902893e-07,
+      "loss": 1.3403,
+      "step": 2555
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3027092671627957e-07,
+      "loss": 1.3742,
+      "step": 2556
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2723669178543414e-07,
+      "loss": 1.5359,
+      "step": 2557
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.242162273173921e-07,
+      "loss": 1.4275,
+      "step": 2558
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2120953761218376e-07,
+      "loss": 1.4927,
+      "step": 2559
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.182166269502307e-07,
+      "loss": 1.6279,
+      "step": 2560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.152374995923413e-07,
+      "loss": 1.4738,
+      "step": 2561
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1227215977969407e-07,
+      "loss": 1.6822,
+      "step": 2562
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.093206117338499e-07,
+      "loss": 1.4295,
+      "step": 2563
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.063828596567242e-07,
+      "loss": 1.3922,
+      "step": 2564
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.034589077306005e-07,
+      "loss": 1.5716,
+      "step": 2565
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.005487601181112e-07,
+      "loss": 1.3622,
+      "step": 2566
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.976524209622367e-07,
+      "loss": 1.4171,
+      "step": 2567
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.947698943863031e-07,
+      "loss": 1.5839,
+      "step": 2568
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9190118449396565e-07,
+      "loss": 1.3866,
+      "step": 2569
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8904629536921856e-07,
+      "loss": 1.3458,
+      "step": 2570
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8620523107637186e-07,
+      "loss": 1.652,
+      "step": 2571
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8337799566005907e-07,
+      "loss": 1.7191,
+      "step": 2572
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.805645931452261e-07,
+      "loss": 1.5986,
+      "step": 2573
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7776502753712243e-07,
+      "loss": 1.5402,
+      "step": 2574
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.749793028213055e-07,
+      "loss": 1.4443,
+      "step": 2575
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7220742296361845e-07,
+      "loss": 1.2336,
+      "step": 2576
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.694493919102059e-07,
+      "loss": 1.4248,
+      "step": 2577
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.667052135874881e-07,
+      "loss": 1.492,
+      "step": 2578
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6397489190216786e-07,
+      "loss": 1.4607,
+      "step": 2579
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6125843074122246e-07,
+      "loss": 1.3069,
+      "step": 2580
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5855583397189077e-07,
+      "loss": 1.2833,
+      "step": 2581
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.558671054416839e-07,
+      "loss": 1.4302,
+      "step": 2582
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.531922489783578e-07,
+      "loss": 1.2546,
+      "step": 2583
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.505312683899297e-07,
+      "loss": 1.5225,
+      "step": 2584
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4788416746465813e-07,
+      "loss": 1.5358,
+      "step": 2585
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.452509499710409e-07,
+      "loss": 1.1396,
+      "step": 2586
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.42631619657816e-07,
+      "loss": 1.3756,
+      "step": 2587
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4002618025394495e-07,
+      "loss": 1.4283,
+      "step": 2588
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3743463546861744e-07,
+      "loss": 1.2582,
+      "step": 2589
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3485698899124333e-07,
+      "loss": 1.4219,
+      "step": 2590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3229324449144165e-07,
+      "loss": 1.5266,
+      "step": 2591
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2974340561904506e-07,
+      "loss": 1.3222,
+      "step": 2592
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2720747600408655e-07,
+      "loss": 1.581,
+      "step": 2593
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2468545925680152e-07,
+      "loss": 1.6482,
+      "step": 2594
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2217735896761128e-07,
+      "loss": 1.3587,
+      "step": 2595
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1968317870713186e-07,
+      "loss": 1.4174,
+      "step": 2596
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1720292202616066e-07,
+      "loss": 1.2659,
+      "step": 2597
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.14736592455671e-07,
+      "loss": 1.4968,
+      "step": 2598
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1228419350681206e-07,
+      "loss": 1.4472,
+      "step": 2599
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0984572867089658e-07,
+      "loss": 1.3207,
+      "step": 2600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0742120141940548e-07,
+      "loss": 1.7412,
+      "step": 2601
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0501061520397547e-07,
+      "loss": 1.4375,
+      "step": 2602
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0261397345639366e-07,
+      "loss": 1.5255,
+      "step": 2603
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0023127958860078e-07,
+      "loss": 1.6913,
+      "step": 2604
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9786253699267567e-07,
+      "loss": 1.4362,
+      "step": 2605
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9550774904083747e-07,
+      "loss": 1.3471,
+      "step": 2606
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9316691908544127e-07,
+      "loss": 1.363,
+      "step": 2607
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9084005045896803e-07,
+      "loss": 1.6219,
+      "step": 2608
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8852714647402571e-07,
+      "loss": 1.548,
+      "step": 2609
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8622821042333927e-07,
+      "loss": 1.2939,
+      "step": 2610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8394324557974962e-07,
+      "loss": 1.4058,
+      "step": 2611
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.816722551962069e-07,
+      "loss": 1.7237,
+      "step": 2612
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7941524250576602e-07,
+      "loss": 1.7418,
+      "step": 2613
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7717221072158786e-07,
+      "loss": 1.4954,
+      "step": 2614
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7494316303692472e-07,
+      "loss": 1.3937,
+      "step": 2615
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.727281026251204e-07,
+      "loss": 1.6045,
+      "step": 2616
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7052703263960912e-07,
+      "loss": 1.2003,
+      "step": 2617
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.683399562139043e-07,
+      "loss": 1.571,
+      "step": 2618
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6616687646160202e-07,
+      "loss": 1.5339,
+      "step": 2619
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6400779647636754e-07,
+      "loss": 1.3924,
+      "step": 2620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6186271933193997e-07,
+      "loss": 1.2853,
+      "step": 2621
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5973164808212094e-07,
+      "loss": 1.3534,
+      "step": 2622
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5761458576077138e-07,
+      "loss": 1.3648,
+      "step": 2623
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5551153538181374e-07,
+      "loss": 1.5253,
+      "step": 2624
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5342249993921532e-07,
+      "loss": 1.4273,
+      "step": 2625
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5134748240700048e-07,
+      "loss": 1.5722,
+      "step": 2626
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4928648573922732e-07,
+      "loss": 1.4696,
+      "step": 2627
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4723951287000214e-07,
+      "loss": 1.5012,
+      "step": 2628
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4520656671346056e-07,
+      "loss": 1.505,
+      "step": 2629
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4318765016377078e-07,
+      "loss": 1.5803,
+      "step": 2630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4118276609513038e-07,
+      "loss": 1.4299,
+      "step": 2631
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.391919173617562e-07,
+      "loss": 1.3572,
+      "step": 2632
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3721510679788774e-07,
+      "loss": 1.3128,
+      "step": 2633
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3525233721777498e-07,
+      "loss": 1.2744,
+      "step": 2634
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3330361141568383e-07,
+      "loss": 1.2885,
+      "step": 2635
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3136893216588175e-07,
+      "loss": 1.2948,
+      "step": 2636
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2944830222264115e-07,
+      "loss": 1.5285,
+      "step": 2637
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2754172432023703e-07,
+      "loss": 1.3903,
+      "step": 2638
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2564920117293266e-07,
+      "loss": 1.3907,
+      "step": 2639
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.237707354749884e-07,
+      "loss": 1.3469,
+      "step": 2640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2190632990064734e-07,
+      "loss": 1.4327,
+      "step": 2641
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2005598710414067e-07,
+      "loss": 1.1509,
+      "step": 2642
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1821970971967579e-07,
+      "loss": 1.2846,
+      "step": 2643
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1639750036143704e-07,
+      "loss": 1.5876,
+      "step": 2644
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.145893616235827e-07,
+      "loss": 1.386,
+      "step": 2645
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1279529608023698e-07,
+      "loss": 1.4907,
+      "step": 2646
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1101530628549128e-07,
+      "loss": 1.4283,
+      "step": 2647
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0924939477339635e-07,
+      "loss": 1.48,
+      "step": 2648
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.074975640579623e-07,
+      "loss": 1.4688,
+      "step": 2649
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0575981663315416e-07,
+      "loss": 1.3441,
+      "step": 2650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.04036154972883e-07,
+      "loss": 1.2524,
+      "step": 2651
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0232658153101261e-07,
+      "loss": 1.5244,
+      "step": 2652
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0063109874134724e-07,
+      "loss": 1.3212,
+      "step": 2653
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.894970901763057e-08,
+      "loss": 1.514,
+      "step": 2654
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.728241475354561e-08,
+      "loss": 1.1713,
+      "step": 2655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.562921832270588e-08,
+      "loss": 1.337,
+      "step": 2656
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.399012207865765e-08,
+      "loss": 1.5239,
+      "step": 2657
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.236512835486989e-08,
+      "loss": 1.5626,
+      "step": 2658
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.075423946473871e-08,
+      "loss": 1.5068,
+      "step": 2659
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.915745770157747e-08,
+      "loss": 1.7189,
+      "step": 2660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.757478533861663e-08,
+      "loss": 1.6673,
+      "step": 2661
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.600622462900165e-08,
+      "loss": 1.3928,
+      "step": 2662
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.445177780578517e-08,
+      "loss": 1.3171,
+      "step": 2663
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.291144708193033e-08,
+      "loss": 1.2642,
+      "step": 2664
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.138523465030191e-08,
+      "loss": 1.3229,
+      "step": 2665
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.98731426836663e-08,
+      "loss": 1.3029,
+      "step": 2666
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.837517333468603e-08,
+      "loss": 1.2242,
+      "step": 2667
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.689132873592076e-08,
+      "loss": 1.6488,
+      "step": 2668
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.542161099981849e-08,
+      "loss": 1.8174,
+      "step": 2669
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.396602221871885e-08,
+      "loss": 1.3012,
+      "step": 2670
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.252456446484534e-08,
+      "loss": 1.409,
+      "step": 2671
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.109723979030536e-08,
+      "loss": 1.4102,
+      "step": 2672
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.968405022708347e-08,
+      "loss": 1.3756,
+      "step": 2673
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.828499778704367e-08,
+      "loss": 1.244,
+      "step": 2674
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.690008446192276e-08,
+      "loss": 1.2074,
+      "step": 2675
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.552931222332803e-08,
+      "loss": 1.4653,
+      "step": 2676
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.417268302273739e-08,
+      "loss": 1.408,
+      "step": 2677
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.283019879149144e-08,
+      "loss": 1.7122,
+      "step": 2678
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.150186144079473e-08,
+      "loss": 1.217,
+      "step": 2679
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.018767286171234e-08,
+      "loss": 1.5713,
+      "step": 2680
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.888763492516436e-08,
+      "loss": 1.4467,
+      "step": 2681
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.760174948193031e-08,
+      "loss": 1.7328,
+      "step": 2682
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.633001836263696e-08,
+      "loss": 1.3985,
+      "step": 2683
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.507244337776274e-08,
+      "loss": 1.6614,
+      "step": 2684
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.382902631763331e-08,
+      "loss": 1.5567,
+      "step": 2685
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.259976895241714e-08,
+      "loss": 1.4546,
+      "step": 2686
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.138467303212546e-08,
+      "loss": 1.6378,
+      "step": 2687
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.018374028660788e-08,
+      "loss": 1.4673,
+      "step": 2688
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.899697242555346e-08,
+      "loss": 1.2791,
+      "step": 2689
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7824371138481815e-08,
+      "loss": 1.4178,
+      "step": 2690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.666593809474762e-08,
+      "loss": 1.4014,
+      "step": 2691
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5521674943534985e-08,
+      "loss": 1.1117,
+      "step": 2692
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.439158331385196e-08,
+      "loss": 1.4885,
+      "step": 2693
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.327566481453715e-08,
+      "loss": 1.2032,
+      "step": 2694
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2173921034246444e-08,
+      "loss": 1.566,
+      "step": 2695
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.108635354145851e-08,
+      "loss": 1.3328,
+      "step": 2696
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.001296388447151e-08,
+      "loss": 1.3768,
+      "step": 2697
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8953753591396415e-08,
+      "loss": 1.5365,
+      "step": 2698
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.790872417016034e-08,
+      "loss": 1.4503,
+      "step": 2699
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.68778771085021e-08,
+      "loss": 1.0304,
+      "step": 2700
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5861213873968904e-08,
+      "loss": 1.6038,
+      "step": 2701
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.48587359139152e-08,
+      "loss": 1.601,
+      "step": 2702
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.38704446555016e-08,
+      "loss": 1.4354,
+      "step": 2703
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.289634150569376e-08,
+      "loss": 1.4827,
+      "step": 2704
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1936427851253503e-08,
+      "loss": 1.0858,
+      "step": 2705
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0990705058748794e-08,
+      "loss": 1.3856,
+      "step": 2706
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0059174474539324e-08,
+      "loss": 1.4821,
+      "step": 2707
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.914183742478427e-08,
+      "loss": 1.5497,
+      "step": 2708
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8238695215432323e-08,
+      "loss": 1.5207,
+      "step": 2709
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.734974913222943e-08,
+      "loss": 1.5781,
+      "step": 2710
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.64750004407055e-08,
+      "loss": 1.6297,
+      "step": 2711
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5614450386182155e-08,
+      "loss": 1.5763,
+      "step": 2712
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4768100193768295e-08,
+      "loss": 1.3805,
+      "step": 2713
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3935951068353446e-08,
+      "loss": 1.5306,
+      "step": 2714
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3118004194614406e-08,
+      "loss": 1.4369,
+      "step": 2715
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2314260737006376e-08,
+      "loss": 1.2034,
+      "step": 2716
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.152472183976406e-08,
+      "loss": 1.5095,
+      "step": 2717
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.074938862690279e-08,
+      "loss": 1.4576,
+      "step": 2718
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.998826220220962e-08,
+      "loss": 1.2408,
+      "step": 2719
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.924134364925112e-08,
+      "loss": 1.4286,
+      "step": 2720
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.850863403136449e-08,
+      "loss": 1.3504,
+      "step": 2721
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7790134391659775e-08,
+      "loss": 1.3752,
+      "step": 2722
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.708584575301542e-08,
+      "loss": 1.4447,
+      "step": 2723
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6395769118080495e-08,
+      "loss": 1.2743,
+      "step": 2724
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.571990546927138e-08,
+      "loss": 1.3814,
+      "step": 2725
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5058255768767295e-08,
+      "loss": 1.3547,
+      "step": 2726
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.441082095851698e-08,
+      "loss": 1.4415,
+      "step": 2727
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3777601960229814e-08,
+      "loss": 1.4448,
+      "step": 2728
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3158599675374695e-08,
+      "loss": 1.4281,
+      "step": 2729
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2553814985186707e-08,
+      "loss": 1.493,
+      "step": 2730
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.19632487506538e-08,
+      "loss": 1.4782,
+      "step": 2731
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1386901812527883e-08,
+      "loss": 1.348,
+      "step": 2732
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0824774991314845e-08,
+      "loss": 1.3377,
+      "step": 2733
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0276869087276764e-08,
+      "loss": 1.4975,
+      "step": 2734
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.74318488042969e-09,
+      "loss": 1.0908,
+      "step": 2735
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.223723130544759e-09,
+      "loss": 1.5807,
+      "step": 2736
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.718484577144859e-09,
+      "loss": 1.3614,
+      "step": 2737
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.227469939503518e-09,
+      "loss": 1.6107,
+      "step": 2738
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.75067991664602e-09,
+      "loss": 1.3086,
+      "step": 2739
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.288115187344957e-09,
+      "loss": 1.4008,
+      "step": 2740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.839776410124677e-09,
+      "loss": 1.4084,
+      "step": 2741
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.4056642232523945e-09,
+      "loss": 1.5999,
+      "step": 2742
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.985779244747081e-09,
+      "loss": 1.732,
+      "step": 2743
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.5801220723683545e-09,
+      "loss": 1.4423,
+      "step": 2744
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.1886932836253675e-09,
+      "loss": 1.5881,
+      "step": 2745
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.811493435766812e-09,
+      "loss": 1.3558,
+      "step": 2746
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.44852306578869e-09,
+      "loss": 1.3489,
+      "step": 2747
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.099782690425435e-09,
+      "loss": 1.2391,
+      "step": 2748
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7652728061576824e-09,
+      "loss": 1.8044,
+      "step": 2749
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.444993889202275e-09,
+      "loss": 1.3535,
+      "step": 2750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1389463955200373e-09,
+      "loss": 1.3158,
+      "step": 2751
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8471307608102238e-09,
+      "loss": 1.4599,
+      "step": 2752
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5695474005116295e-09,
+      "loss": 1.2495,
+      "step": 2753
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3061967097992575e-09,
+      "loss": 1.6066,
+      "step": 2754
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.057079063589873e-09,
+      "loss": 1.5093,
+      "step": 2755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8221948165342285e-09,
+      "loss": 1.4025,
+      "step": 2756
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6015443030215073e-09,
+      "loss": 1.3788,
+      "step": 2757
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3951278371759913e-09,
+      "loss": 1.4646,
+      "step": 2758
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2029457128615029e-09,
+      "loss": 1.353,
+      "step": 2759
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0249982036725226e-09,
+      "loss": 1.4964,
+      "step": 2760
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.612855629419603e-10,
+      "loss": 1.3455,
+      "step": 2761
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.11808023735605e-10,
+      "loss": 1.7719,
+      "step": 2762
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.765657988554551e-10,
+      "loss": 1.4145,
+      "step": 2763
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.555590808374977e-10,
+      "loss": 1.4484,
+      "step": 2764
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.4878804194948825e-10,
+      "loss": 1.3246,
+      "step": 2765
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.562528341942816e-10,
+      "loss": 1.4381,
+      "step": 2766
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.779535893076112e-10,
+      "loss": 1.2841,
+      "step": 2767
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1389041876030959e-10,
+      "loss": 1.1594,
+      "step": 2768
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.406341375497782e-11,
+      "loss": 1.278,
+      "step": 2769
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.847264522487514e-11,
+      "loss": 1.2287,
+      "step": 2770
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.118163839470171e-12,
+      "loss": 1.2849,
+      "step": 2771
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 1.6482,
+      "step": 2772
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2772,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 3.786131489826013e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}