diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13795 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9968171060148534,
+  "global_step": 2295,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8985507246376816e-07,
+      "loss": 1.4613,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.797101449275363e-07,
+      "loss": 1.4612,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 1.2091,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1594202898550726e-06,
+      "loss": 1.1679,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4492753623188408e-06,
+      "loss": 1.1503,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 1.1265,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.028985507246377e-06,
+      "loss": 1.1165,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3188405797101453e-06,
+      "loss": 1.0889,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 1.0583,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8985507246376816e-06,
+      "loss": 1.0624,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.188405797101449e-06,
+      "loss": 1.0271,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 0.9918,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.768115942028986e-06,
+      "loss": 1.0008,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.057971014492754e-06,
+      "loss": 0.9744,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 0.9572,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637681159420291e-06,
+      "loss": 0.9621,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.927536231884059e-06,
+      "loss": 0.9357,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 0.9194,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.507246376811595e-06,
+      "loss": 0.9199,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.797101449275363e-06,
+      "loss": 0.9156,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 0.8953,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.376811594202898e-06,
+      "loss": 0.8773,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.8721,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 0.8576,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.246376811594203e-06,
+      "loss": 0.878,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.536231884057972e-06,
+      "loss": 0.8563,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 0.8681,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.115942028985508e-06,
+      "loss": 0.8587,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.405797101449275e-06,
+      "loss": 0.8605,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 0.8414,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.985507246376812e-06,
+      "loss": 0.831,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.275362318840581e-06,
+      "loss": 0.8234,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 0.8351,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.855072463768118e-06,
+      "loss": 0.8443,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0144927536231885e-05,
+      "loss": 0.842,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 0.8201,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0724637681159422e-05,
+      "loss": 0.826,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.101449275362319e-05,
+      "loss": 0.8138,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 0.8232,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1594202898550726e-05,
+      "loss": 0.8128,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1884057971014494e-05,
+      "loss": 0.8167,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 0.8092,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2463768115942029e-05,
+      "loss": 0.824,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2753623188405797e-05,
+      "loss": 0.8109,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 0.8065,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.803,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3623188405797103e-05,
+      "loss": 0.812,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 0.8128,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.420289855072464e-05,
+      "loss": 0.7964,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4492753623188407e-05,
+      "loss": 0.8103,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 0.8032,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5072463768115944e-05,
+      "loss": 0.7852,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.536231884057971e-05,
+      "loss": 0.8033,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 0.7928,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5942028985507246e-05,
+      "loss": 0.7979,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6231884057971015e-05,
+      "loss": 0.7901,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 0.7749,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.681159420289855e-05,
+      "loss": 0.7997,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.710144927536232e-05,
+      "loss": 0.7827,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 0.8002,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7681159420289858e-05,
+      "loss": 0.7926,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7971014492753624e-05,
+      "loss": 0.7883,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 0.7936,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8550724637681162e-05,
+      "loss": 0.7939,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8840579710144928e-05,
+      "loss": 0.7863,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 0.7828,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9420289855072467e-05,
+      "loss": 0.7858,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9710144927536236e-05,
+      "loss": 0.7728,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.7885,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999004091687e-05,
+      "loss": 0.7766,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999960163687307e-05,
+      "loss": 0.7768,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999910368370826e-05,
+      "loss": 0.7819,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999840655066608e-05,
+      "loss": 0.7804,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999751023913506e-05,
+      "loss": 0.7718,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999964147509006e-05,
+      "loss": 0.775,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999951200881446e-05,
+      "loss": 0.7838,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999362625344584e-05,
+      "loss": 0.7751,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999193324977974e-05,
+      "loss": 0.7835,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999004108051846e-05,
+      "loss": 0.784,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998794974943087e-05,
+      "loss": 0.7617,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998565926068253e-05,
+      "loss": 0.7698,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998316961883563e-05,
+      "loss": 0.7688,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999804808288491e-05,
+      "loss": 0.7694,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997759289607854e-05,
+      "loss": 0.7812,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997450582627614e-05,
+      "loss": 0.7768,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999712196255908e-05,
+      "loss": 0.7702,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996773430056806e-05,
+      "loss": 0.7874,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996404985814998e-05,
+      "loss": 0.7691,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996016630567535e-05,
+      "loss": 0.7719,
+      "step": 89
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995608365087945e-05,
+      "loss": 0.7675,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995180190189424e-05,
+      "loss": 0.7635,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999473210672481e-05,
+      "loss": 0.7627,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999426411558661e-05,
+      "loss": 0.7644,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993776217706972e-05,
+      "loss": 0.7655,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993268414057704e-05,
+      "loss": 0.7627,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992740705650252e-05,
+      "loss": 0.762,
+      "step": 96
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999219309353572e-05,
+      "loss": 0.7663,
+      "step": 97
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999162557880485e-05,
+      "loss": 0.7552,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991038162588018e-05,
+      "loss": 0.7698,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999043084605526e-05,
+      "loss": 0.7596,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998980363041624e-05,
+      "loss": 0.7657,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989156516920248e-05,
+      "loss": 0.7826,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988489506856218e-05,
+      "loss": 0.7664,
+      "step": 103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987802601552717e-05,
+      "loss": 0.7646,
+      "step": 104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987095802377933e-05,
+      "loss": 0.7726,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986369110739675e-05,
+      "loss": 0.7574,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985622528085382e-05,
+      "loss": 0.7565,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998485605590211e-05,
+      "loss": 0.7617,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984069695716534e-05,
+      "loss": 0.7676,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998326344909494e-05,
+      "loss": 0.7598,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982437317643218e-05,
+      "loss": 0.7565,
+      "step": 111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998159130300687e-05,
+      "loss": 0.7483,
+      "step": 112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980725406871007e-05,
+      "loss": 0.7509,
+      "step": 113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979839630960333e-05,
+      "loss": 0.7674,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997893397703915e-05,
+      "loss": 0.7576,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978008446911354e-05,
+      "loss": 0.7646,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977063042420438e-05,
+      "loss": 0.7611,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976097765449463e-05,
+      "loss": 0.7532,
+      "step": 118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975112617921097e-05,
+      "loss": 0.7571,
+      "step": 119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997410760179756e-05,
+      "loss": 0.755,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973082719080673e-05,
+      "loss": 0.7448,
+      "step": 121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972037971811802e-05,
+      "loss": 0.7534,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99709733620719e-05,
+      "loss": 0.7553,
+      "step": 123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969888891981473e-05,
+      "loss": 0.7383,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968784563700586e-05,
+      "loss": 0.7569,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967660379428855e-05,
+      "loss": 0.7551,
+      "step": 126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966516341405452e-05,
+      "loss": 0.7606,
+      "step": 127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996535245190909e-05,
+      "loss": 0.7511,
+      "step": 128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996416871325803e-05,
+      "loss": 0.756,
+      "step": 129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996296512781005e-05,
+      "loss": 0.7435,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996174169796248e-05,
+      "loss": 0.7613,
+      "step": 131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996049842615217e-05,
+      "loss": 0.7478,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959235314855485e-05,
+      "loss": 0.7709,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9957952366588307e-05,
+      "loss": 0.759,
+      "step": 134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995664958390604e-05,
+      "loss": 0.7439,
+      "step": 135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955326969403587e-05,
+      "loss": 0.7577,
+      "step": 136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953984525715354e-05,
+      "loss": 0.7415,
+      "step": 137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995262225551524e-05,
+      "loss": 0.7583,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951240161516643e-05,
+      "loss": 0.7458,
+      "step": 139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949838246472436e-05,
+      "loss": 0.7346,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948416513174976e-05,
+      "loss": 0.7505,
+      "step": 141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946974964456094e-05,
+      "loss": 0.7518,
+      "step": 142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945513603187096e-05,
+      "loss": 0.7364,
+      "step": 143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944032432278743e-05,
+      "loss": 0.7379,
+      "step": 144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942531454681254e-05,
+      "loss": 0.7418,
+      "step": 145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941010673384307e-05,
+      "loss": 0.7538,
+      "step": 146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939470091417012e-05,
+      "loss": 0.749,
+      "step": 147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993790971184793e-05,
+      "loss": 0.7411,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936329537785054e-05,
+      "loss": 0.746,
+      "step": 149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934729572375792e-05,
+      "loss": 0.7505,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993310981880699e-05,
+      "loss": 0.7323,
+      "step": 151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931470280304895e-05,
+      "loss": 0.7503,
+      "step": 152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992981096013517e-05,
+      "loss": 0.7414,
+      "step": 153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928131861602878e-05,
+      "loss": 0.7398,
+      "step": 154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992643298805247e-05,
+      "loss": 0.7419,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992471434286779e-05,
+      "loss": 0.7442,
+      "step": 156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922975929472076e-05,
+      "loss": 0.7415,
+      "step": 157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921217751327916e-05,
+      "loss": 0.752,
+      "step": 158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919439811937283e-05,
+      "loss": 0.7352,
+      "step": 159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917642114841505e-05,
+      "loss": 0.7351,
+      "step": 160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915824663621267e-05,
+      "loss": 0.739,
+      "step": 161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913987461896597e-05,
+      "loss": 0.7465,
+      "step": 162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912130513326863e-05,
+      "loss": 0.7476,
+      "step": 163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991025382161077e-05,
+      "loss": 0.7491,
+      "step": 164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908357390486342e-05,
+      "loss": 0.7532,
+      "step": 165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990644122373092e-05,
+      "loss": 0.7364,
+      "step": 166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990450532516116e-05,
+      "loss": 0.7263,
+      "step": 167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990254969863301e-05,
+      "loss": 0.748,
+      "step": 168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900574348041728e-05,
+      "loss": 0.7482,
+      "step": 169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898579277321845e-05,
+      "loss": 0.7449,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989656449044718e-05,
+      "loss": 0.7207,
+      "step": 171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894529991430814e-05,
+      "loss": 0.7382,
+      "step": 172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98924757843251e-05,
+      "loss": 0.7457,
+      "step": 173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890401873221642e-05,
+      "loss": 0.7394,
+      "step": 174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 0.7414,
+      "step": 175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886194955584126e-05,
+      "loss": 0.7293,
+      "step": 176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988406195742948e-05,
+      "loss": 0.7438,
+      "step": 177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881909272035888e-05,
+      "loss": 0.7334,
+      "step": 178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879736903691107e-05,
+      "loss": 0.7381,
+      "step": 179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877544856722094e-05,
+      "loss": 0.737,
+      "step": 180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875333135495e-05,
+      "loss": 0.731,
+      "step": 181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987310174441518e-05,
+      "loss": 0.7525,
+      "step": 182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987085068792715e-05,
+      "loss": 0.7439,
+      "step": 183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98685799705146e-05,
+      "loss": 0.7396,
+      "step": 184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866289596700383e-05,
+      "loss": 0.7349,
+      "step": 185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98639795710465e-05,
+      "loss": 0.7382,
+      "step": 186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861649898154107e-05,
+      "loss": 0.7386,
+      "step": 187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985930058266348e-05,
+      "loss": 0.73,
+      "step": 188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856931629254032e-05,
+      "loss": 0.733,
+      "step": 189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985454304264427e-05,
+      "loss": 0.7427,
+      "step": 190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985213482759183e-05,
+      "loss": 0.727,
+      "step": 191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849706988893433e-05,
+      "loss": 0.7265,
+      "step": 192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984725953138489e-05,
+      "loss": 0.7308,
+      "step": 193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984479245994108e-05,
+      "loss": 0.7382,
+      "step": 194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984230577947597e-05,
+      "loss": 0.7495,
+      "step": 195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983979949494256e-05,
+      "loss": 0.7387,
+      "step": 196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837273611332918e-05,
+      "loss": 0.728,
+      "step": 197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983472813367813e-05,
+      "loss": 0.7352,
+      "step": 198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832163067048335e-05,
+      "loss": 0.747,
+      "step": 199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982957841655266e-05,
+      "loss": 0.7307,
+      "step": 200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826974187339267e-05,
+      "loss": 0.7248,
+      "step": 201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824350384595295e-05,
+      "loss": 0.7386,
+      "step": 202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821707013546885e-05,
+      "loss": 0.7282,
+      "step": 203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819044079459142e-05,
+      "loss": 0.7402,
+      "step": 204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816361587636143e-05,
+      "loss": 0.7422,
+      "step": 205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813659543420924e-05,
+      "loss": 0.7233,
+      "step": 206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981093795219546e-05,
+      "loss": 0.7477,
+      "step": 207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808196819380656e-05,
+      "loss": 0.7375,
+      "step": 208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805436150436352e-05,
+      "loss": 0.7331,
+      "step": 209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980265595086129e-05,
+      "loss": 0.7415,
+      "step": 210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799856226193125e-05,
+      "loss": 0.728,
+      "step": 211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797036982008385e-05,
+      "loss": 0.7192,
+      "step": 212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794198223922496e-05,
+      "loss": 0.7334,
+      "step": 213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979133995758974e-05,
+      "loss": 0.7311,
+      "step": 214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978846218870326e-05,
+      "loss": 0.7414,
+      "step": 215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785564922995042e-05,
+      "loss": 0.7285,
+      "step": 216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978264816623591e-05,
+      "loss": 0.7268,
+      "step": 217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977971192423551e-05,
+      "loss": 0.7292,
+      "step": 218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776756202842297e-05,
+      "loss": 0.7295,
+      "step": 219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977378100794352e-05,
+      "loss": 0.727,
+      "step": 220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977078634546523e-05,
+      "loss": 0.7543,
+      "step": 221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976777222137224e-05,
+      "loss": 0.7277,
+      "step": 222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764738641668137e-05,
+      "loss": 0.7307,
+      "step": 223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761685612395253e-05,
+      "loss": 0.7171,
+      "step": 224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758613139634662e-05,
+      "loss": 0.7222,
+      "step": 225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755521229506164e-05,
+      "loss": 0.7243,
+      "step": 226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752409888168285e-05,
+      "loss": 0.7247,
+      "step": 227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.7336,
+      "step": 228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974612893669194e-05,
+      "loss": 0.7282,
+      "step": 229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742959339063977e-05,
+      "loss": 0.7285,
+      "step": 230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739770335247616e-05,
+      "loss": 0.7371,
+      "step": 231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973656193159476e-05,
+      "loss": 0.7338,
+      "step": 232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733334134495963e-05,
+      "loss": 0.7328,
+      "step": 233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730086950380404e-05,
+      "loss": 0.7358,
+      "step": 234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726820385715877e-05,
+      "loss": 0.726,
+      "step": 235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723534447008785e-05,
+      "loss": 0.7413,
+      "step": 236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972022914080411e-05,
+      "loss": 0.7388,
+      "step": 237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716904473685417e-05,
+      "loss": 0.7217,
+      "step": 238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971356045227484e-05,
+      "loss": 0.7273,
+      "step": 239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710197083233044e-05,
+      "loss": 0.7171,
+      "step": 240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970681437325925e-05,
+      "loss": 0.724,
+      "step": 241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97034123290912e-05,
+      "loss": 0.7371,
+      "step": 242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699990957505136e-05,
+      "loss": 0.7196,
+      "step": 243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696550265315805e-05,
+      "loss": 0.723,
+      "step": 244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693090259376436e-05,
+      "loss": 0.7333,
+      "step": 245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689610946578725e-05,
+      "loss": 0.7305,
+      "step": 246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686112333852826e-05,
+      "loss": 0.7225,
+      "step": 247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968259442816733e-05,
+      "loss": 0.7224,
+      "step": 248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679057236529266e-05,
+      "loss": 0.7187,
+      "step": 249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675500765984066e-05,
+      "loss": 0.7265,
+      "step": 250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671925023615572e-05,
+      "loss": 0.7109,
+      "step": 251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668330016546004e-05,
+      "loss": 0.7307,
+      "step": 252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664715751935958e-05,
+      "loss": 0.7303,
+      "step": 253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661082236984387e-05,
+      "loss": 0.733,
+      "step": 254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965742947892858e-05,
+      "loss": 0.7299,
+      "step": 255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965375748504417e-05,
+      "loss": 0.7212,
+      "step": 256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650066262645097e-05,
+      "loss": 0.7246,
+      "step": 257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964635581908359e-05,
+      "loss": 0.7255,
+      "step": 258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642626161750176e-05,
+      "loss": 0.7137,
+      "step": 259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638877298073645e-05,
+      "loss": 0.7074,
+      "step": 260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635109235521057e-05,
+      "loss": 0.7266,
+      "step": 261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963132198159769e-05,
+      "loss": 0.7386,
+      "step": 262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627515543847068e-05,
+      "loss": 0.7255,
+      "step": 263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962368992985091e-05,
+      "loss": 0.7244,
+      "step": 264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961984514722914e-05,
+      "loss": 0.7311,
+      "step": 265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961598120363986e-05,
+      "loss": 0.7149,
+      "step": 266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961209810677934e-05,
+      "loss": 0.723,
+      "step": 267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608195864381994e-05,
+      "loss": 0.7423,
+      "step": 268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960427448422037e-05,
+      "loss": 0.7295,
+      "step": 269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600333974105145e-05,
+      "loss": 0.7218,
+      "step": 270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596374341885093e-05,
+      "loss": 0.7306,
+      "step": 271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592395595447064e-05,
+      "loss": 0.7459,
+      "step": 272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588397742716004e-05,
+      "loss": 0.7351,
+      "step": 273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584380791654896e-05,
+      "loss": 0.7152,
+      "step": 274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958034475026477e-05,
+      "loss": 0.7268,
+      "step": 275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576289626584685e-05,
+      "loss": 0.7167,
+      "step": 276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95722154286917e-05,
+      "loss": 0.7381,
+      "step": 277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568122164700867e-05,
+      "loss": 0.7289,
+      "step": 278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564009842765225e-05,
+      "loss": 0.7164,
+      "step": 279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559878471075763e-05,
+      "loss": 0.7312,
+      "step": 280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.7207,
+      "step": 281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955155861138903e-05,
+      "loss": 0.7317,
+      "step": 282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547370139963406e-05,
+      "loss": 0.718,
+      "step": 283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954316265192719e-05,
+      "loss": 0.7216,
+      "step": 284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538936155660934e-05,
+      "loss": 0.716,
+      "step": 285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534690659583045e-05,
+      "loss": 0.7168,
+      "step": 286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953042617214977e-05,
+      "loss": 0.7296,
+      "step": 287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526142701855187e-05,
+      "loss": 0.7191,
+      "step": 288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521840257231183e-05,
+      "loss": 0.7282,
+      "step": 289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517518846847437e-05,
+      "loss": 0.71,
+      "step": 290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951317847931141e-05,
+      "loss": 0.7314,
+      "step": 291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508819163268315e-05,
+      "loss": 0.722,
+      "step": 292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504440907401113e-05,
+      "loss": 0.711,
+      "step": 293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500043720430484e-05,
+      "loss": 0.7228,
+      "step": 294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495627611114817e-05,
+      "loss": 0.7395,
+      "step": 295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491192588250198e-05,
+      "loss": 0.732,
+      "step": 296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486738660670373e-05,
+      "loss": 0.7226,
+      "step": 297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948226583724675e-05,
+      "loss": 0.7238,
+      "step": 298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947777412688838e-05,
+      "loss": 0.6971,
+      "step": 299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473263538541916e-05,
+      "loss": 0.7199,
+      "step": 300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468734081191627e-05,
+      "loss": 0.7145,
+      "step": 301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464185763859365e-05,
+      "loss": 0.7198,
+      "step": 302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945961859560454e-05,
+      "loss": 0.7157,
+      "step": 303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455032585524115e-05,
+      "loss": 0.7088,
+      "step": 304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450427742752583e-05,
+      "loss": 0.7294,
+      "step": 305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445804076461946e-05,
+      "loss": 0.7152,
+      "step": 306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944116159586169e-05,
+      "loss": 0.7243,
+      "step": 307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94365003101988e-05,
+      "loss": 0.7213,
+      "step": 308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943182022875769e-05,
+      "loss": 0.7205,
+      "step": 309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9427121360860233e-05,
+      "loss": 0.7258,
+      "step": 310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422403715865708e-05,
+      "loss": 0.7272,
+      "step": 311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417667303170803e-05,
+      "loss": 0.7098,
+      "step": 312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412912132209573e-05,
+      "loss": 0.7136,
+      "step": 313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408138212453456e-05,
+      "loss": 0.7182,
+      "step": 314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940334555341122e-05,
+      "loss": 0.7207,
+      "step": 315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398534164628968e-05,
+      "loss": 0.7225,
+      "step": 316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93937040556901e-05,
+      "loss": 0.7198,
+      "step": 317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938885523621531e-05,
+      "loss": 0.707,
+      "step": 318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383987715862554e-05,
+      "loss": 0.7167,
+      "step": 319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379101504327044e-05,
+      "loss": 0.7214,
+      "step": 320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374196611341212e-05,
+      "loss": 0.7206,
+      "step": 321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369273046674708e-05,
+      "loss": 0.7093,
+      "step": 322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936433082013437e-05,
+      "loss": 0.7097,
+      "step": 323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935936994156421e-05,
+      "loss": 0.7312,
+      "step": 324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354390420845387e-05,
+      "loss": 0.704,
+      "step": 325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934939226789619e-05,
+      "loss": 0.7264,
+      "step": 326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344375492672024e-05,
+      "loss": 0.7225,
+      "step": 327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933934010516539e-05,
+      "loss": 0.7223,
+      "step": 328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933428611540585e-05,
+      "loss": 0.7223,
+      "step": 329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932921353346003e-05,
+      "loss": 0.7249,
+      "step": 330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932412236943158e-05,
+      "loss": 0.7144,
+      "step": 331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319012633461164e-05,
+      "loss": 0.7053,
+      "step": 332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9313884335726443e-05,
+      "loss": 0.7125,
+      "step": 333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308737486442045e-05,
+      "loss": 0.7301,
+      "step": 334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303572095859545e-05,
+      "loss": 0.7082,
+      "step": 335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298388174267463e-05,
+      "loss": 0.7083,
+      "step": 336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293185731991212e-05,
+      "loss": 0.7165,
+      "step": 337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287964779393105e-05,
+      "loss": 0.7061,
+      "step": 338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282725326872324e-05,
+      "loss": 0.7109,
+      "step": 339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92774673848649e-05,
+      "loss": 0.7212,
+      "step": 340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927219096384368e-05,
+      "loss": 0.7139,
+      "step": 341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266896074318335e-05,
+      "loss": 0.7098,
+      "step": 342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261582726835316e-05,
+      "loss": 0.7272,
+      "step": 343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925625093197783e-05,
+      "loss": 0.7171,
+      "step": 344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250900700365837e-05,
+      "loss": 0.7118,
+      "step": 345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924553204265602e-05,
+      "loss": 0.708,
+      "step": 346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240144969541754e-05,
+      "loss": 0.7294,
+      "step": 347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923473949175311e-05,
+      "loss": 0.719,
+      "step": 348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229315620056805e-05,
+      "loss": 0.7258,
+      "step": 349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223873365256192e-05,
+      "loss": 0.7139,
+      "step": 350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921841273819125e-05,
+      "loss": 0.7216,
+      "step": 351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212933749738546e-05,
+      "loss": 0.718,
+      "step": 352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920743641081122e-05,
+      "loss": 0.7218,
+      "step": 353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920192073235896e-05,
+      "loss": 0.7141,
+      "step": 354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919638672536799e-05,
+      "loss": 0.7089,
+      "step": 355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9190834400861035e-05,
+      "loss": 0.7128,
+      "step": 356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918526376989731e-05,
+      "loss": 0.724,
+      "step": 357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179674843572487e-05,
+      "loss": 0.7303,
+      "step": 358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174067633018682e-05,
+      "loss": 0.7188,
+      "step": 359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168442149404438e-05,
+      "loss": 0.7218,
+      "step": 360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916279840393467e-05,
+      "loss": 0.7226,
+      "step": 361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157136407850706e-05,
+      "loss": 0.7139,
+      "step": 362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151456172430186e-05,
+      "loss": 0.7169,
+      "step": 363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145757708987103e-05,
+      "loss": 0.7344,
+      "step": 364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914004102887176e-05,
+      "loss": 0.7155,
+      "step": 365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134306143470722e-05,
+      "loss": 0.7122,
+      "step": 366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128553064206835e-05,
+      "loss": 0.7308,
+      "step": 367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912278180253918e-05,
+      "loss": 0.7177,
+      "step": 368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911699236996305e-05,
+      "loss": 0.7161,
+      "step": 369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111184778009934e-05,
+      "loss": 0.7202,
+      "step": 370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105359038247484e-05,
+      "loss": 0.7148,
+      "step": 371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099515162279515e-05,
+      "loss": 0.7339,
+      "step": 372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909365316174595e-05,
+      "loss": 0.717,
+      "step": 373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908777304832282e-05,
+      "loss": 0.7112,
+      "step": 374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081874833722234e-05,
+      "loss": 0.7141,
+      "step": 375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075958529692355e-05,
+      "loss": 0.7291,
+      "step": 376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070024148017375e-05,
+      "loss": 0.7214,
+      "step": 377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906407170051749e-05,
+      "loss": 0.7043,
+      "step": 378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.905810119904889e-05,
+      "loss": 0.7137,
+      "step": 379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052112655503713e-05,
+      "loss": 0.7191,
+      "step": 380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046106081810047e-05,
+      "loss": 0.721,
+      "step": 381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904008148993188e-05,
+      "loss": 0.723,
+      "step": 382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903403889186909e-05,
+      "loss": 0.7262,
+      "step": 383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027978299657436e-05,
+      "loss": 0.7176,
+      "step": 384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021899725368498e-05,
+      "loss": 0.7122,
+      "step": 385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015803181109685e-05,
+      "loss": 0.701,
+      "step": 386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.7039,
+      "step": 387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003556231290988e-05,
+      "loss": 0.7299,
+      "step": 388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8997405850124786e-05,
+      "loss": 0.7101,
+      "step": 389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991237547776014e-05,
+      "loss": 0.7135,
+      "step": 390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89850513365308e-05,
+      "loss": 0.7123,
+      "step": 391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897884722871094e-05,
+      "loss": 0.7043,
+      "step": 392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972625236673887e-05,
+      "loss": 0.6995,
+      "step": 393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966385372812703e-05,
+      "loss": 0.7103,
+      "step": 394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896012764955605e-05,
+      "loss": 0.7239,
+      "step": 395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895385207936817e-05,
+      "loss": 0.7182,
+      "step": 396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947558674748844e-05,
+      "loss": 0.716,
+      "step": 397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8941247448233386e-05,
+      "loss": 0.7159,
+      "step": 398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934918412392596e-05,
+      "loss": 0.7052,
+      "step": 399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928571579832756e-05,
+      "loss": 0.7204,
+      "step": 400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892220696319559e-05,
+      "loss": 0.719,
+      "step": 401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891582457515825e-05,
+      "loss": 0.7114,
+      "step": 402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909424428433278e-05,
+      "loss": 0.715,
+      "step": 403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.89030065357686e-05,
+      "loss": 0.7034,
+      "step": 404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896570909947477e-05,
+      "loss": 0.7126,
+      "step": 405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889011756378849e-05,
+      "loss": 0.71,
+      "step": 406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888364651014553e-05,
+      "loss": 0.7181,
+      "step": 407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887715776190775e-05,
+      "loss": 0.6989,
+      "step": 408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870651331999542e-05,
+      "loss": 0.715,
+      "step": 409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886412723338052e-05,
+      "loss": 0.7087,
+      "step": 410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857585479045493e-05,
+      "loss": 0.694,
+      "step": 411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885102608202444e-05,
+      "loss": 0.7022,
+      "step": 412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844449055382473e-05,
+      "loss": 0.7047,
+      "step": 413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837854412219828e-05,
+      "loss": 0.7088,
+      "step": 414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831242165671816e-05,
+      "loss": 0.6996,
+      "step": 415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824612328908828e-05,
+      "loss": 0.7097,
+      "step": 416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8817964915136277e-05,
+      "loss": 0.7021,
+      "step": 417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811299937594598e-05,
+      "loss": 0.7057,
+      "step": 418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88046174095592e-05,
+      "loss": 0.7119,
+      "step": 419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797917344340453e-05,
+      "loss": 0.7064,
+      "step": 420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791199755283664e-05,
+      "loss": 0.7015,
+      "step": 421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784464655769033e-05,
+      "loss": 0.7115,
+      "step": 422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8777712059211643e-05,
+      "loss": 0.7099,
+      "step": 423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770941979061432e-05,
+      "loss": 0.7067,
+      "step": 424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8764154428803155e-05,
+      "loss": 0.7159,
+      "step": 425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875734942195637e-05,
+      "loss": 0.7152,
+      "step": 426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87505269720754e-05,
+      "loss": 0.7134,
+      "step": 427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743687092749318e-05,
+      "loss": 0.7153,
+      "step": 428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736829797601903e-05,
+      "loss": 0.7159,
+      "step": 429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729955100291634e-05,
+      "loss": 0.7071,
+      "step": 430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872306301451165e-05,
+      "loss": 0.7119,
+      "step": 431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716153553989716e-05,
+      "loss": 0.7024,
+      "step": 432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8709226732488216e-05,
+      "loss": 0.7011,
+      "step": 433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.87022825638041e-05,
+      "loss": 0.7066,
+      "step": 434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695321061768886e-05,
+      "loss": 0.7107,
+      "step": 435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868834224024861e-05,
+      "loss": 0.6861,
+      "step": 436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86813461131438e-05,
+      "loss": 0.7031,
+      "step": 437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674332694389464e-05,
+      "loss": 0.7129,
+      "step": 438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8667301997955038e-05,
+      "loss": 0.6993,
+      "step": 439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.7201,
+      "step": 440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653188828095754e-05,
+      "loss": 0.7089,
+      "step": 441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646106382781738e-05,
+      "loss": 0.7056,
+      "step": 442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8639006716009275e-05,
+      "loss": 0.6948,
+      "step": 443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631889841919596e-05,
+      "loss": 0.705,
+      "step": 444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862475577468821e-05,
+      "loss": 0.7129,
+      "step": 445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617604528524876e-05,
+      "loss": 0.7113,
+      "step": 446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610436117673557e-05,
+      "loss": 0.7052,
+      "step": 447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8603250556412418e-05,
+      "loss": 0.6966,
+      "step": 448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596047859053776e-05,
+      "loss": 0.6996,
+      "step": 449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588828039944086e-05,
+      "loss": 0.7143,
+      "step": 450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8581591113463903e-05,
+      "loss": 0.7119,
+      "step": 451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8574337094027858e-05,
+      "loss": 0.6901,
+      "step": 452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567065996084628e-05,
+      "loss": 0.7012,
+      "step": 453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8559777834116906e-05,
+      "loss": 0.7152,
+      "step": 454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552472622641372e-05,
+      "loss": 0.703,
+      "step": 455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854515037620867e-05,
+      "loss": 0.6992,
+      "step": 456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537811109403372e-05,
+      "loss": 0.7109,
+      "step": 457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530454836843953e-05,
+      "loss": 0.7022,
+      "step": 458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8523081573182754e-05,
+      "loss": 0.7163,
+      "step": 459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851569133310597e-05,
+      "loss": 0.7051,
+      "step": 460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508284131333604e-05,
+      "loss": 0.7185,
+      "step": 461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500859982619438e-05,
+      "loss": 0.7177,
+      "step": 462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8493418901751016e-05,
+      "loss": 0.7129,
+      "step": 463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485960903549614e-05,
+      "loss": 0.6994,
+      "step": 464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847848600287019e-05,
+      "loss": 0.7091,
+      "step": 465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470994214601378e-05,
+      "loss": 0.7025,
+      "step": 466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846348555366544e-05,
+      "loss": 0.6966,
+      "step": 467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845596003501826e-05,
+      "loss": 0.7065,
+      "step": 468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448417673649292e-05,
+      "loss": 0.7114,
+      "step": 469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844085848458153e-05,
+      "loss": 0.7066,
+      "step": 470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433282482871497e-05,
+      "loss": 0.7096,
+      "step": 471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425689683609198e-05,
+      "loss": 0.7107,
+      "step": 472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418080101918095e-05,
+      "loss": 0.7045,
+      "step": 473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841045375295508e-05,
+      "loss": 0.7029,
+      "step": 474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402810651910444e-05,
+      "loss": 0.714,
+      "step": 475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839515081400784e-05,
+      "loss": 0.7024,
+      "step": 476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387474254504265e-05,
+      "loss": 0.7117,
+      "step": 477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8379780988690014e-05,
+      "loss": 0.708,
+      "step": 478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837207103188866e-05,
+      "loss": 0.7083,
+      "step": 479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364344399457028e-05,
+      "loss": 0.6947,
+      "step": 480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356601106785148e-05,
+      "loss": 0.6935,
+      "step": 481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8348841169296247e-05,
+      "loss": 0.7018,
+      "step": 482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341064602446686e-05,
+      "loss": 0.7063,
+      "step": 483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333271421725968e-05,
+      "loss": 0.6916,
+      "step": 484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325461642656676e-05,
+      "loss": 0.7189,
+      "step": 485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8317635280794466e-05,
+      "loss": 0.7059,
+      "step": 486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309792351728006e-05,
+      "loss": 0.7107,
+      "step": 487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301932871078975e-05,
+      "loss": 0.7095,
+      "step": 488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829405685450202e-05,
+      "loss": 0.6945,
+      "step": 489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828616431768472e-05,
+      "loss": 0.7085,
+      "step": 490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8278255276347563e-05,
+      "loss": 0.7019,
+      "step": 491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270329746243903e-05,
+      "loss": 0.7004,
+      "step": 492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.6993,
+      "step": 493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254429282914715e-05,
+      "loss": 0.6858,
+      "step": 494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824645438135999e-05,
+      "loss": 0.7157,
+      "step": 495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823846305438032e-05,
+      "loss": 0.7022,
+      "step": 496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8230455317892957e-05,
+      "loss": 0.7063,
+      "step": 497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822243118784785e-05,
+      "loss": 0.7113,
+      "step": 498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8214390680227588e-05,
+      "loss": 0.6959,
+      "step": 499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82063338110474e-05,
+      "loss": 0.6957,
+      "step": 500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198260596355077e-05,
+      "loss": 0.7066,
+      "step": 501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819017105223099e-05,
+      "loss": 0.7018,
+      "step": 502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182065194788024e-05,
+      "loss": 0.6978,
+      "step": 503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8173943040171567e-05,
+      "loss": 0.7046,
+      "step": 504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8165804604559455e-05,
+      "loss": 0.715,
+      "step": 505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157649904161963e-05,
+      "loss": 0.7014,
+      "step": 506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814947895522176e-05,
+      "loss": 0.6954,
+      "step": 507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141291774013876e-05,
+      "loss": 0.7016,
+      "step": 508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8133088376845675e-05,
+      "loss": 0.7081,
+      "step": 509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124868780056814e-05,
+      "loss": 0.7126,
+      "step": 510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116633000019233e-05,
+      "loss": 0.7173,
+      "step": 511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108381053137086e-05,
+      "loss": 0.7183,
+      "step": 512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100112955846746e-05,
+      "loss": 0.7145,
+      "step": 513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8091828724616742e-05,
+      "loss": 0.7048,
+      "step": 514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083528375947744e-05,
+      "loss": 0.6982,
+      "step": 515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807521192637253e-05,
+      "loss": 0.7033,
+      "step": 516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8066879392455932e-05,
+      "loss": 0.71,
+      "step": 517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058530790794837e-05,
+      "loss": 0.695,
+      "step": 518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805016613801813e-05,
+      "loss": 0.7,
+      "step": 519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8041785450786655e-05,
+      "loss": 0.7003,
+      "step": 520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033388745793218e-05,
+      "loss": 0.7041,
+      "step": 521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8024976039762507e-05,
+      "loss": 0.6964,
+      "step": 522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801654734945109e-05,
+      "loss": 0.7016,
+      "step": 523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800810269164738e-05,
+      "loss": 0.7177,
+      "step": 524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999642083171576e-05,
+      "loss": 0.7057,
+      "step": 525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7991165540875662e-05,
+      "loss": 0.6979,
+      "step": 526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7982673081643364e-05,
+      "loss": 0.7034,
+      "step": 527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974164722390094e-05,
+      "loss": 0.7132,
+      "step": 528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7965640480062945e-05,
+      "loss": 0.7173,
+      "step": 529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795710037164065e-05,
+      "loss": 0.7058,
+      "step": 530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7948544414133534e-05,
+      "loss": 0.7149,
+      "step": 531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7939972624583496e-05,
+      "loss": 0.6997,
+      "step": 532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793138502006397e-05,
+      "loss": 0.7031,
+      "step": 533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792278161767989e-05,
+      "loss": 0.7066,
+      "step": 534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914162434567653e-05,
+      "loss": 0.7006,
+      "step": 535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905527487895095e-05,
+      "loss": 0.7021,
+      "step": 536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7896876794861443e-05,
+      "loss": 0.731,
+      "step": 537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7888210372697292e-05,
+      "loss": 0.703,
+      "step": 538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879528238664567e-05,
+      "loss": 0.7081,
+      "step": 539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870830410056487e-05,
+      "loss": 0.71,
+      "step": 540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862116904197534e-05,
+      "loss": 0.7114,
+      "step": 541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7853387738443407e-05,
+      "loss": 0.7029,
+      "step": 542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844642930181008e-05,
+      "loss": 0.6932,
+      "step": 543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7835882496828392e-05,
+      "loss": 0.6917,
+      "step": 544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782710645583473e-05,
+      "loss": 0.7047,
+      "step": 545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.6933,
+      "step": 546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7809507620876406e-05,
+      "loss": 0.697,
+      "step": 547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800684861965386e-05,
+      "loss": 0.6983,
+      "step": 548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779184656552056e-05,
+      "loss": 0.7075,
+      "step": 549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778299274914619e-05,
+      "loss": 0.7201,
+      "step": 550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774123430477464e-05,
+      "loss": 0.691,
+      "step": 551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765238627180424e-05,
+      "loss": 0.7039,
+      "step": 552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775633835695198e-05,
+      "loss": 0.6932,
+      "step": 553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7747422637519838e-05,
+      "loss": 0.7096,
+      "step": 554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.773849148664247e-05,
+      "loss": 0.7039,
+      "step": 555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729544922109097e-05,
+      "loss": 0.6931,
+      "step": 556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7720582961739628e-05,
+      "loss": 0.7031,
+      "step": 557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711605623384653e-05,
+      "loss": 0.7129,
+      "step": 558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702612924925377e-05,
+      "loss": 0.7007,
+      "step": 559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769360488427361e-05,
+      "loss": 0.7042,
+      "step": 560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684581519371714e-05,
+      "loss": 0.6943,
+      "step": 561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767554284819258e-05,
+      "loss": 0.6895,
+      "step": 562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666488888739587e-05,
+      "loss": 0.7025,
+      "step": 563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657419659046555e-05,
+      "loss": 0.6974,
+      "step": 564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7648335177177725e-05,
+      "loss": 0.7106,
+      "step": 565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639235461227727e-05,
+      "loss": 0.7007,
+      "step": 566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630120529321518e-05,
+      "loss": 0.7055,
+      "step": 567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620990399614385e-05,
+      "loss": 0.6967,
+      "step": 568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7611845090291858e-05,
+      "loss": 0.6996,
+      "step": 569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760268461956972e-05,
+      "loss": 0.6991,
+      "step": 570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759350900569395e-05,
+      "loss": 0.699,
+      "step": 571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758431826694069e-05,
+      "loss": 0.7103,
+      "step": 572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575112421616203e-05,
+      "loss": 0.7056,
+      "step": 573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7565891488056846e-05,
+      "loss": 0.7066,
+      "step": 574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7556655484629028e-05,
+      "loss": 0.7,
+      "step": 575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547404429729174e-05,
+      "loss": 0.7056,
+      "step": 576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753813834178369e-05,
+      "loss": 0.705,
+      "step": 577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7528857239248925e-05,
+      "loss": 0.7104,
+      "step": 578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751956114061113e-05,
+      "loss": 0.7103,
+      "step": 579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751025006438643e-05,
+      "loss": 0.6979,
+      "step": 580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7500924029120782e-05,
+      "loss": 0.6991,
+      "step": 581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7491583053389937e-05,
+      "loss": 0.7117,
+      "step": 582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748222715579941e-05,
+      "loss": 0.7098,
+      "step": 583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7472856354984427e-05,
+      "loss": 0.7045,
+      "step": 584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7463470669609907e-05,
+      "loss": 0.6988,
+      "step": 585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7454070118370416e-05,
+      "loss": 0.7019,
+      "step": 586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7444654719990128e-05,
+      "loss": 0.6855,
+      "step": 587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743522449322279e-05,
+      "loss": 0.6997,
+      "step": 588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7425779456851683e-05,
+      "loss": 0.6916,
+      "step": 589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741631962968959e-05,
+      "loss": 0.6787,
+      "step": 590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406845030578747e-05,
+      "loss": 0.6952,
+      "step": 591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397355678390823e-05,
+      "loss": 0.6956,
+      "step": 592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7387851592026868e-05,
+      "loss": 0.7023,
+      "step": 593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378332790417275e-05,
+      "loss": 0.6989,
+      "step": 594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368799292521754e-05,
+      "loss": 0.704,
+      "step": 595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359251117329284e-05,
+      "loss": 0.7099,
+      "step": 596
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734968828385808e-05,
+      "loss": 0.7002,
+      "step": 597
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734011081115555e-05,
+      "loss": 0.6858,
+      "step": 598
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.6971,
+      "step": 599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7320912024391915e-05,
+      "loss": 0.695,
+      "step": 600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7311290748571273e-05,
+      "loss": 0.6794,
+      "step": 601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730165491000016e-05,
+      "loss": 0.6908,
+      "step": 602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729200452787139e-05,
+      "loss": 0.7013,
+      "step": 603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7282339621406764e-05,
+      "loss": 0.7041,
+      "step": 604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7272660209857e-05,
+      "loss": 0.6884,
+      "step": 605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726296631250171e-05,
+      "loss": 0.6938,
+      "step": 606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7253257948649357e-05,
+      "loss": 0.7078,
+      "step": 607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243535137637227e-05,
+      "loss": 0.7031,
+      "step": 608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233797898831376e-05,
+      "loss": 0.6978,
+      "step": 609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224046251626596e-05,
+      "loss": 0.6871,
+      "step": 610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721428021544638e-05,
+      "loss": 0.6931,
+      "step": 611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720449980974288e-05,
+      "loss": 0.6917,
+      "step": 612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7194705053996873e-05,
+      "loss": 0.6939,
+      "step": 613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7184895967717715e-05,
+      "loss": 0.7031,
+      "step": 614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717507257044331e-05,
+      "loss": 0.7074,
+      "step": 615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7165234881740064e-05,
+      "loss": 0.6998,
+      "step": 616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7155382921202844e-05,
+      "loss": 0.7031,
+      "step": 617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7145516708454953e-05,
+      "loss": 0.6927,
+      "step": 618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713563626314808e-05,
+      "loss": 0.7052,
+      "step": 619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7125741604962254e-05,
+      "loss": 0.6997,
+      "step": 620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711583275360582e-05,
+      "loss": 0.6938,
+      "step": 621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7105909728815398e-05,
+      "loss": 0.7114,
+      "step": 622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709597255035583e-05,
+      "loss": 0.6836,
+      "step": 623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7086021238020154e-05,
+      "loss": 0.6879,
+      "step": 624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076055811629556e-05,
+      "loss": 0.6898,
+      "step": 625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066076291033348e-05,
+      "loss": 0.7005,
+      "step": 626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056082696108896e-05,
+      "loss": 0.7124,
+      "step": 627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7046075046761614e-05,
+      "loss": 0.6931,
+      "step": 628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036053362924896e-05,
+      "loss": 0.6984,
+      "step": 629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026017664560108e-05,
+      "loss": 0.684,
+      "step": 630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7015967971656513e-05,
+      "loss": 0.6898,
+      "step": 631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7005904304231263e-05,
+      "loss": 0.6955,
+      "step": 632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699582668232934e-05,
+      "loss": 0.7061,
+      "step": 633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985735126023505e-05,
+      "loss": 0.6958,
+      "step": 634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6975629655414304e-05,
+      "loss": 0.678,
+      "step": 635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965510290629973e-05,
+      "loss": 0.6962,
+      "step": 636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695537705182643e-05,
+      "loss": 0.6864,
+      "step": 637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6945229959187232e-05,
+      "loss": 0.69,
+      "step": 638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6935069032923525e-05,
+      "loss": 0.6958,
+      "step": 639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6924894293274016e-05,
+      "loss": 0.7031,
+      "step": 640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6914705760504913e-05,
+      "loss": 0.7006,
+      "step": 641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904503454909905e-05,
+      "loss": 0.6974,
+      "step": 642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689428739681012e-05,
+      "loss": 0.7009,
+      "step": 643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688405760655407e-05,
+      "loss": 0.6998,
+      "step": 644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6873814104517617e-05,
+      "loss": 0.6961,
+      "step": 645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6863556911103948e-05,
+      "loss": 0.6972,
+      "step": 646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6853286046743505e-05,
+      "loss": 0.6952,
+      "step": 647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843001531893967e-05,
+      "loss": 0.6918,
+      "step": 648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68327033870402e-05,
+      "loss": 0.6997,
+      "step": 649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682239163269422e-05,
+      "loss": 0.7079,
+      "step": 650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6812066289395157e-05,
+      "loss": 0.7018,
+      "step": 651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6801727377709195e-05,
+      "loss": 0.6917,
+      "step": 652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679137491822955e-05,
+      "loss": 0.7007,
+      "step": 653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6781008931576433e-05,
+      "loss": 0.6916,
+      "step": 654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6770629438396973e-05,
+      "loss": 0.698,
+      "step": 655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676023645936523e-05,
+      "loss": 0.7022,
+      "step": 656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749830015182106e-05,
+      "loss": 0.7069,
+      "step": 657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673941012657533e-05,
+      "loss": 0.6931,
+      "step": 658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6728976814299413e-05,
+      "loss": 0.7047,
+      "step": 659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718530099135596e-05,
+      "loss": 0.699,
+      "step": 660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670807000189182e-05,
+      "loss": 0.6857,
+      "step": 661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6697596543402684e-05,
+      "loss": 0.6911,
+      "step": 662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687109744529394e-05,
+      "loss": 0.6856,
+      "step": 663
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667660962615973e-05,
+      "loss": 0.682,
+      "step": 664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6666096209208e-05,
+      "loss": 0.6875,
+      "step": 665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665556951461501e-05,
+      "loss": 0.702,
+      "step": 666
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6645029563348e-05,
+      "loss": 0.694,
+      "step": 667
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6634476376400624e-05,
+      "loss": 0.6968,
+      "step": 668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6623909974792888e-05,
+      "loss": 0.6921,
+      "step": 669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6613330379571135e-05,
+      "loss": 0.7008,
+      "step": 670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6602737611807975e-05,
+      "loss": 0.708,
+      "step": 671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6592131692602257e-05,
+      "loss": 0.6878,
+      "step": 672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6581512643079028e-05,
+      "loss": 0.6904,
+      "step": 673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657088048438949e-05,
+      "loss": 0.6948,
+      "step": 674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656023523771095e-05,
+      "loss": 0.699,
+      "step": 675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654957692424679e-05,
+      "loss": 0.6964,
+      "step": 676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6538905565226416e-05,
+      "loss": 0.7015,
+      "step": 677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528221181905217e-05,
+      "loss": 0.6847,
+      "step": 678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517523795564527e-05,
+      "loss": 0.6927,
+      "step": 679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6506813427511574e-05,
+      "loss": 0.6926,
+      "step": 680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496090099079452e-05,
+      "loss": 0.6996,
+      "step": 681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648535383162706e-05,
+      "loss": 0.7003,
+      "step": 682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647460464653908e-05,
+      "loss": 0.6991,
+      "step": 683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6463842565225914e-05,
+      "loss": 0.6852,
+      "step": 684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6453067609123656e-05,
+      "loss": 0.6982,
+      "step": 685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442279799694037e-05,
+      "loss": 0.6916,
+      "step": 686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64314791584244e-05,
+      "loss": 0.6861,
+      "step": 687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6420665706827647e-05,
+      "loss": 0.6949,
+      "step": 688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640983946644218e-05,
+      "loss": 0.6923,
+      "step": 689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399000458831893e-05,
+      "loss": 0.6848,
+      "step": 690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388148705586097e-05,
+      "loss": 0.6974,
+      "step": 691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377284228319496e-05,
+      "loss": 0.6927,
+      "step": 692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6366407048672135e-05,
+      "loss": 0.7069,
+      "step": 693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355517188309366e-05,
+      "loss": 0.6961,
+      "step": 694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6344614668921787e-05,
+      "loss": 0.6904,
+      "step": 695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6333699512225224e-05,
+      "loss": 0.6929,
+      "step": 696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6322771739960664e-05,
+      "loss": 0.6948,
+      "step": 697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6311831373894228e-05,
+      "loss": 0.6977,
+      "step": 698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300878435817115e-05,
+      "loss": 0.6832,
+      "step": 699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.628991294754557e-05,
+      "loss": 0.685,
+      "step": 700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278934930920834e-05,
+      "loss": 0.6977,
+      "step": 701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626794440780911e-05,
+      "loss": 0.6973,
+      "step": 702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62569414001015e-05,
+      "loss": 0.6851,
+      "step": 703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245925929713976e-05,
+      "loss": 0.7001,
+      "step": 704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.6864,
+      "step": 705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622385768868716e-05,
+      "loss": 0.692,
+      "step": 706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6212804962003757e-05,
+      "loss": 0.6896,
+      "step": 707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620173986055213e-05,
+      "loss": 0.6997,
+      "step": 708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6190662406371937e-05,
+      "loss": 0.702,
+      "step": 709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6179572621527434e-05,
+      "loss": 0.6949,
+      "step": 710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616847052810744e-05,
+      "loss": 0.6837,
+      "step": 711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6157356148225282e-05,
+      "loss": 0.6955,
+      "step": 712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6146229504018777e-05,
+      "loss": 0.6903,
+      "step": 713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613509061765015e-05,
+      "loss": 0.6971,
+      "step": 714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6123939511306028e-05,
+      "loss": 0.6925,
+      "step": 715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6112776207197368e-05,
+      "loss": 0.689,
+      "step": 716
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101600727559423e-05,
+      "loss": 0.6876,
+      "step": 717
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60904130946517e-05,
+      "loss": 0.6859,
+      "step": 718
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607921333075791e-05,
+      "loss": 0.6891,
+      "step": 719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068001458185934e-05,
+      "loss": 0.6955,
+      "step": 720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6056777499267764e-05,
+      "loss": 0.7016,
+      "step": 721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6045541476359464e-05,
+      "loss": 0.6951,
+      "step": 722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603429341184114e-05,
+      "loss": 0.702,
+      "step": 723
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6023033328116863e-05,
+      "loss": 0.696,
+      "step": 724
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6011761247614664e-05,
+      "loss": 0.6966,
+      "step": 725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600047719278645e-05,
+      "loss": 0.6983,
+      "step": 726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5989181186108003e-05,
+      "loss": 0.6914,
+      "step": 727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597787325007889e-05,
+      "loss": 0.6817,
+      "step": 728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596655340722244e-05,
+      "loss": 0.6894,
+      "step": 729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5955221680085715e-05,
+      "loss": 0.6829,
+      "step": 730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594387809123943e-05,
+      "loss": 0.698,
+      "step": 731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593252266327794e-05,
+      "loss": 0.7011,
+      "step": 732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592115541881917e-05,
+      "loss": 0.6882,
+      "step": 733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5909776380504583e-05,
+      "loss": 0.6964,
+      "step": 734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5898385570999146e-05,
+      "loss": 0.6901,
+      "step": 735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588698301299126e-05,
+      "loss": 0.6955,
+      "step": 736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5875568729192728e-05,
+      "loss": 0.6957,
+      "step": 737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5864142742338707e-05,
+      "loss": 0.694,
+      "step": 738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5852705075187674e-05,
+      "loss": 0.6942,
+      "step": 739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584125575052136e-05,
+      "loss": 0.7015,
+      "step": 740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829794791144723e-05,
+      "loss": 0.6922,
+      "step": 741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5818322219885888e-05,
+      "loss": 0.6911,
+      "step": 742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580683805959612e-05,
+      "loss": 0.6888,
+      "step": 743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5795342333149757e-05,
+      "loss": 0.6915,
+      "step": 744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5783835063444176e-05,
+      "loss": 0.6948,
+      "step": 745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577231627339975e-05,
+      "loss": 0.7015,
+      "step": 746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57607859859598e-05,
+      "loss": 0.6904,
+      "step": 747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5749244224090537e-05,
+      "loss": 0.6899,
+      "step": 748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573769101078104e-05,
+      "loss": 0.6915,
+      "step": 749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572612636904319e-05,
+      "loss": 0.6945,
+      "step": 750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5714550321911636e-05,
+      "loss": 0.6983,
+      "step": 751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570296289244373e-05,
+      "loss": 0.6975,
+      "step": 752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691364103719515e-05,
+      "loss": 0.6958,
+      "step": 753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5679753978841655e-05,
+      "loss": 0.6894,
+      "step": 754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566813254093538e-05,
+      "loss": 0.7002,
+      "step": 755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565649981314847e-05,
+      "loss": 0.6854,
+      "step": 756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5644855818651184e-05,
+      "loss": 0.6855,
+      "step": 757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563320058063622e-05,
+      "loss": 0.6984,
+      "step": 758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5621534122318682e-05,
+      "loss": 0.7037,
+      "step": 759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5609856466936013e-05,
+      "loss": 0.7027,
+      "step": 760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559816763774796e-05,
+      "loss": 0.6895,
+      "step": 761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586467658036526e-05,
+      "loss": 0.6926,
+      "step": 762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5574756551105926e-05,
+      "loss": 0.6951,
+      "step": 763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556303434028254e-05,
+      "loss": 0.693,
+      "step": 764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5551301048914863e-05,
+      "loss": 0.6955,
+      "step": 765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5539556700373454e-05,
+      "loss": 0.6855,
+      "step": 766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5527801318050904e-05,
+      "loss": 0.6398,
+      "step": 767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5516034925361785e-05,
+      "loss": 0.6401,
+      "step": 768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5504257545742585e-05,
+      "loss": 0.6437,
+      "step": 769
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492469202651687e-05,
+      "loss": 0.6335,
+      "step": 770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5480669919569313e-05,
+      "loss": 0.6198,
+      "step": 771
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5468859719997468e-05,
+      "loss": 0.6241,
+      "step": 772
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5457038627459905e-05,
+      "loss": 0.6261,
+      "step": 773
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544520666550207e-05,
+      "loss": 0.6409,
+      "step": 774
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5433363857691067e-05,
+      "loss": 0.6192,
+      "step": 775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5421510227615587e-05,
+      "loss": 0.649,
+      "step": 776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54096457988859e-05,
+      "loss": 0.6335,
+      "step": 777
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5397770595133763e-05,
+      "loss": 0.6349,
+      "step": 778
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538588464001241e-05,
+      "loss": 0.6442,
+      "step": 779
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537398795719648e-05,
+      "loss": 0.636,
+      "step": 780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536208057038199e-05,
+      "loss": 0.6289,
+      "step": 781
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5350162503286262e-05,
+      "loss": 0.6266,
+      "step": 782
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533823377964791e-05,
+      "loss": 0.63,
+      "step": 783
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5326294423226754e-05,
+      "loss": 0.6413,
+      "step": 784
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314344457803812e-05,
+      "loss": 0.621,
+      "step": 785
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5302383907181212e-05,
+      "loss": 0.6376,
+      "step": 786
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290412795182193e-05,
+      "loss": 0.6241,
+      "step": 787
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5278431145651006e-05,
+      "loss": 0.6332,
+      "step": 788
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52664389824529e-05,
+      "loss": 0.6273,
+      "step": 789
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254436329474062e-05,
+      "loss": 0.6366,
+      "step": 790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5242423210621584e-05,
+      "loss": 0.6267,
+      "step": 791
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5230399649823389e-05,
+      "loss": 0.6308,
+      "step": 792
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5218365671028207e-05,
+      "loss": 0.6217,
+      "step": 793
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.520632129820552e-05,
+      "loss": 0.6213,
+      "step": 794
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194266555345505e-05,
+      "loss": 0.6305,
+      "step": 795
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5182201466459007e-05,
+      "loss": 0.6343,
+      "step": 796
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517012605557746e-05,
+      "loss": 0.6252,
+      "step": 797
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158040346752878e-05,
+      "loss": 0.6292,
+      "step": 798
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514594436405777e-05,
+      "loss": 0.6379,
+      "step": 799
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.513383813158512e-05,
+      "loss": 0.6369,
+      "step": 800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121721673448319e-05,
+      "loss": 0.6286,
+      "step": 801
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109595013781136e-05,
+      "loss": 0.6381,
+      "step": 802
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097458176737647e-05,
+      "loss": 0.6354,
+      "step": 803
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5085311186492206e-05,
+      "loss": 0.6275,
+      "step": 804
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5073154067239396e-05,
+      "loss": 0.6374,
+      "step": 805
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506098684319396e-05,
+      "loss": 0.6374,
+      "step": 806
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5048809538590789e-05,
+      "loss": 0.624,
+      "step": 807
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5036622177684832e-05,
+      "loss": 0.6352,
+      "step": 808
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5024424784751079e-05,
+      "loss": 0.6243,
+      "step": 809
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5012217384084501e-05,
+      "loss": 0.6305,
+      "step": 810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6348,
+      "step": 811
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4987772656832369e-05,
+      "loss": 0.6229,
+      "step": 812
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975535378936228e-05,
+      "loss": 0.6283,
+      "step": 813
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4963288190685993e-05,
+      "loss": 0.6307,
+      "step": 814
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4951031116475819e-05,
+      "loss": 0.6421,
+      "step": 815
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938764180719548e-05,
+      "loss": 0.6365,
+      "step": 816
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4926487407850667e-05,
+      "loss": 0.6338,
+      "step": 817
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914200822322255e-05,
+      "loss": 0.6312,
+      "step": 818
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490190444860694e-05,
+      "loss": 0.6314,
+      "step": 819
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488959831119684e-05,
+      "loss": 0.6217,
+      "step": 820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4877282434603527e-05,
+      "loss": 0.624,
+      "step": 821
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4864956843357967e-05,
+      "loss": 0.6456,
+      "step": 822
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4852621562010484e-05,
+      "loss": 0.6388,
+      "step": 823
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4840276615130687e-05,
+      "loss": 0.6246,
+      "step": 824
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482792202730745e-05,
+      "loss": 0.631,
+      "step": 825
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481555782314885e-05,
+      "loss": 0.6302,
+      "step": 826
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480318402728211e-05,
+      "loss": 0.6256,
+      "step": 827
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790800664353562e-05,
+      "loss": 0.6508,
+      "step": 828
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4778407759028599e-05,
+      "loss": 0.6344,
+      "step": 829
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4766005335991608e-05,
+      "loss": 0.6279,
+      "step": 830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475359341994595e-05,
+      "loss": 0.6294,
+      "step": 831
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4741172035613884e-05,
+      "loss": 0.626,
+      "step": 832
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728741207736525e-05,
+      "loss": 0.6299,
+      "step": 833
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716300961073808e-05,
+      "loss": 0.6298,
+      "step": 834
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4703851320404416e-05,
+      "loss": 0.637,
+      "step": 835
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691392310525761e-05,
+      "loss": 0.6228,
+      "step": 836
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4678923956253894e-05,
+      "loss": 0.627,
+      "step": 837
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.46664462824235e-05,
+      "loss": 0.6443,
+      "step": 838
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4653959313887813e-05,
+      "loss": 0.6299,
+      "step": 839
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4641463075518587e-05,
+      "loss": 0.6342,
+      "step": 840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462895759220603e-05,
+      "loss": 0.6316,
+      "step": 841
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4616442888858781e-05,
+      "loss": 0.6367,
+      "step": 842
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460391899040383e-05,
+      "loss": 0.6229,
+      "step": 843
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4591385921786484e-05,
+      "loss": 0.6379,
+      "step": 844
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4578843707970323e-05,
+      "loss": 0.6461,
+      "step": 845
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4566292373937133e-05,
+      "loss": 0.6276,
+      "step": 846
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455373194468687e-05,
+      "loss": 0.6274,
+      "step": 847
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4541162445237604e-05,
+      "loss": 0.6329,
+      "step": 848
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528583900625481e-05,
+      "loss": 0.6361,
+      "step": 849
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4515996335904646e-05,
+      "loss": 0.6249,
+      "step": 850
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4503399776147223e-05,
+      "loss": 0.6396,
+      "step": 851
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490794246443249e-05,
+      "loss": 0.6238,
+      "step": 852
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4478179771900634e-05,
+      "loss": 0.629,
+      "step": 853
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4465556377645086e-05,
+      "loss": 0.6266,
+      "step": 854
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4452924088820101e-05,
+      "loss": 0.6375,
+      "step": 855
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444028293058688e-05,
+      "loss": 0.6274,
+      "step": 856
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4427632928124288e-05,
+      "loss": 0.6355,
+      "step": 857
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4414974106628816e-05,
+      "loss": 0.6385,
+      "step": 858
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4402306491314508e-05,
+      "loss": 0.6422,
+      "step": 859
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389630107412942e-05,
+      "loss": 0.6344,
+      "step": 860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4376944980173138e-05,
+      "loss": 0.6298,
+      "step": 861
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364251134861552e-05,
+      "loss": 0.6348,
+      "step": 862
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435154859676199e-05,
+      "loss": 0.6344,
+      "step": 863
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.6374,
+      "step": 864
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432611754342072e-05,
+      "loss": 0.6326,
+      "step": 865
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313389078833005e-05,
+      "loss": 0.6366,
+      "step": 866
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4300652022765207e-05,
+      "loss": 0.645,
+      "step": 867
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4287906400587204e-05,
+      "loss": 0.6257,
+      "step": 868
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4275152237685938e-05,
+      "loss": 0.6342,
+      "step": 869
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4262389559465364e-05,
+      "loss": 0.6283,
+      "step": 870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4249618391346399e-05,
+      "loss": 0.6359,
+      "step": 871
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423683875876686e-05,
+      "loss": 0.6391,
+      "step": 872
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4224050687181442e-05,
+      "loss": 0.6146,
+      "step": 873
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211254202061632e-05,
+      "loss": 0.6372,
+      "step": 874
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198449328895685e-05,
+      "loss": 0.6415,
+      "step": 875
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4185636093188558e-05,
+      "loss": 0.6375,
+      "step": 876
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172814520461867e-05,
+      "loss": 0.6288,
+      "step": 877
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4159984636253834e-05,
+      "loss": 0.6379,
+      "step": 878
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147146466119235e-05,
+      "loss": 0.6367,
+      "step": 879
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134300035629353e-05,
+      "loss": 0.6414,
+      "step": 880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4121445370371922e-05,
+      "loss": 0.6363,
+      "step": 881
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4108582495951077e-05,
+      "loss": 0.6304,
+      "step": 882
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4095711437987303e-05,
+      "loss": 0.6266,
+      "step": 883
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4082832222117388e-05,
+      "loss": 0.6275,
+      "step": 884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406994487399437e-05,
+      "loss": 0.6279,
+      "step": 885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4057049419287486e-05,
+      "loss": 0.6319,
+      "step": 886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4044145883682108e-05,
+      "loss": 0.6354,
+      "step": 887
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4031234292879726e-05,
+      "loss": 0.6218,
+      "step": 888
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018314672597848e-05,
+      "loss": 0.6235,
+      "step": 889
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005387048569997e-05,
+      "loss": 0.6267,
+      "step": 890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3992451446545624e-05,
+      "loss": 0.6282,
+      "step": 891
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3979507892290078e-05,
+      "loss": 0.6303,
+      "step": 892
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3966556411584548e-05,
+      "loss": 0.6411,
+      "step": 893
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953597030226006e-05,
+      "loss": 0.6272,
+      "step": 894
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394062977402717e-05,
+      "loss": 0.6245,
+      "step": 895
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927654668816425e-05,
+      "loss": 0.6357,
+      "step": 896
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3914671740437811e-05,
+      "loss": 0.6379,
+      "step": 897
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3901681014750932e-05,
+      "loss": 0.6302,
+      "step": 898
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.388868251763094e-05,
+      "loss": 0.6287,
+      "step": 899
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875676274968453e-05,
+      "loss": 0.6246,
+      "step": 900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3862662312669518e-05,
+      "loss": 0.6293,
+      "step": 901
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3849640656655572e-05,
+      "loss": 0.6288,
+      "step": 902
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3836611332863356e-05,
+      "loss": 0.6244,
+      "step": 903
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3823574367244898e-05,
+      "loss": 0.6401,
+      "step": 904
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3810529785767444e-05,
+      "loss": 0.636,
+      "step": 905
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3797477614413404e-05,
+      "loss": 0.6358,
+      "step": 906
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3784417879180314e-05,
+      "loss": 0.6388,
+      "step": 907
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377135060608077e-05,
+      "loss": 0.6395,
+      "step": 908
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3758275821142382e-05,
+      "loss": 0.6216,
+      "step": 909
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374519355040773e-05,
+      "loss": 0.6216,
+      "step": 910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373210381993429e-05,
+      "loss": 0.6289,
+      "step": 911
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3719006655794414e-05,
+      "loss": 0.635,
+      "step": 912
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3705902084075244e-05,
+      "loss": 0.6377,
+      "step": 913
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3692790130878684e-05,
+      "loss": 0.6388,
+      "step": 914
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3679670822321347e-05,
+      "loss": 0.6196,
+      "step": 915
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3666544184534483e-05,
+      "loss": 0.6183,
+      "step": 916
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.6294,
+      "step": 917
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640269025870152e-05,
+      "loss": 0.6361,
+      "step": 918
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3627120557327982e-05,
+      "loss": 0.6326,
+      "step": 919
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3613964864226778e-05,
+      "loss": 0.6399,
+      "step": 920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3600801972770272e-05,
+      "loss": 0.6382,
+      "step": 921
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3587631909176526e-05,
+      "loss": 0.6367,
+      "step": 922
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574454699677893e-05,
+      "loss": 0.615,
+      "step": 923
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561270370520957e-05,
+      "loss": 0.6318,
+      "step": 924
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3548078947966487e-05,
+      "loss": 0.6361,
+      "step": 925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3534880458289377e-05,
+      "loss": 0.6387,
+      "step": 926
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3521674927778594e-05,
+      "loss": 0.6353,
+      "step": 927
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350846238273714e-05,
+      "loss": 0.6354,
+      "step": 928
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3495242849481973e-05,
+      "loss": 0.6284,
+      "step": 929
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348201635434399e-05,
+      "loss": 0.62,
+      "step": 930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3468782923667936e-05,
+      "loss": 0.6336,
+      "step": 931
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345554258381238e-05,
+      "loss": 0.6396,
+      "step": 932
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3442295361149651e-05,
+      "loss": 0.6239,
+      "step": 933
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3429041282065788e-05,
+      "loss": 0.627,
+      "step": 934
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341578037296049e-05,
+      "loss": 0.6321,
+      "step": 935
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402512660247046e-05,
+      "loss": 0.6463,
+      "step": 936
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3389238170352318e-05,
+      "loss": 0.6286,
+      "step": 937
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.337595692971665e-05,
+      "loss": 0.6296,
+      "step": 938
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336266896479384e-05,
+      "loss": 0.6416,
+      "step": 939
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349374302051072e-05,
+      "loss": 0.6271,
+      "step": 940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336072967968882e-05,
+      "loss": 0.6278,
+      "step": 941
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322764989041086e-05,
+      "loss": 0.6248,
+      "step": 942
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330945039177474e-05,
+      "loss": 0.6239,
+      "step": 943
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.329612920269008e-05,
+      "loss": 0.6294,
+      "step": 944
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.328280144832047e-05,
+      "loss": 0.6196,
+      "step": 945
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3269467155212352e-05,
+      "loss": 0.6402,
+      "step": 946
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256126349925195e-05,
+      "loss": 0.6331,
+      "step": 947
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3242779059031436e-05,
+      "loss": 0.6347,
+      "step": 948
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322942530911643e-05,
+      "loss": 0.637,
+      "step": 949
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216065126778399e-05,
+      "loss": 0.6287,
+      "step": 950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3202698538628376e-05,
+      "loss": 0.637,
+      "step": 951
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3189325571290158e-05,
+      "loss": 0.6384,
+      "step": 952
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3175946251400234e-05,
+      "loss": 0.6253,
+      "step": 953
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162560605607763e-05,
+      "loss": 0.6351,
+      "step": 954
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3149168660574495e-05,
+      "loss": 0.6247,
+      "step": 955
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313577044297473e-05,
+      "loss": 0.6307,
+      "step": 956
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3122365979495259e-05,
+      "loss": 0.6354,
+      "step": 957
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108955296835313e-05,
+      "loss": 0.6349,
+      "step": 958
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309553842170652e-05,
+      "loss": 0.6325,
+      "step": 959
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082115380832829e-05,
+      "loss": 0.6327,
+      "step": 960
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3068686200950475e-05,
+      "loss": 0.6414,
+      "step": 961
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055250908807926e-05,
+      "loss": 0.6249,
+      "step": 962
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041809531165819e-05,
+      "loss": 0.6327,
+      "step": 963
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3028362094796908e-05,
+      "loss": 0.6338,
+      "step": 964
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3014908626486032e-05,
+      "loss": 0.6435,
+      "step": 965
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3001449153030024e-05,
+      "loss": 0.6265,
+      "step": 966
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2987983701237688e-05,
+      "loss": 0.6299,
+      "step": 967
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2974512297929733e-05,
+      "loss": 0.611,
+      "step": 968
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2961034969938732e-05,
+      "loss": 0.6337,
+      "step": 969
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2947551744109044e-05,
+      "loss": 0.6299,
+      "step": 970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934062647296783e-05,
+      "loss": 0.6306,
+      "step": 971
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292056770636976e-05,
+      "loss": 0.642,
+      "step": 972
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290706694820742e-05,
+      "loss": 0.65,
+      "step": 973
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2893560399700797e-05,
+      "loss": 0.6337,
+      "step": 974
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288004808775246e-05,
+      "loss": 0.6333,
+      "step": 975
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2866530039276455e-05,
+      "loss": 0.6162,
+      "step": 976
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2853006281198257e-05,
+      "loss": 0.6319,
+      "step": 977
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2839476840454714e-05,
+      "loss": 0.6397,
+      "step": 978
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282594174399399e-05,
+      "loss": 0.6421,
+      "step": 979
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2812401018775512e-05,
+      "loss": 0.6354,
+      "step": 980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798854691769927e-05,
+      "loss": 0.6145,
+      "step": 981
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278530278995903e-05,
+      "loss": 0.637,
+      "step": 982
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2771745340335726e-05,
+      "loss": 0.6321,
+      "step": 983
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2758182369903969e-05,
+      "loss": 0.6299,
+      "step": 984
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2744613905678707e-05,
+      "loss": 0.6256,
+      "step": 985
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2731039974685833e-05,
+      "loss": 0.6393,
+      "step": 986
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717460603962132e-05,
+      "loss": 0.6192,
+      "step": 987
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2703875820555218e-05,
+      "loss": 0.6263,
+      "step": 988
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269028565152349e-05,
+      "loss": 0.636,
+      "step": 989
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2676690123936066e-05,
+      "loss": 0.6347,
+      "step": 990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2663089264872751e-05,
+      "loss": 0.6299,
+      "step": 991
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2649483101423962e-05,
+      "loss": 0.6347,
+      "step": 992
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2635871660690677e-05,
+      "loss": 0.6379,
+      "step": 993
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2622254969784393e-05,
+      "loss": 0.628,
+      "step": 994
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2608633055827064e-05,
+      "loss": 0.6283,
+      "step": 995
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259500594595104e-05,
+      "loss": 0.6304,
+      "step": 996
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2581373667299026e-05,
+      "loss": 0.6383,
+      "step": 997
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2567736247024023e-05,
+      "loss": 0.6339,
+      "step": 998
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2554093712289267e-05,
+      "loss": 0.6246,
+      "step": 999
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2540446090268193e-05,
+      "loss": 0.6246,
+      "step": 1000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2526793408144355e-05,
+      "loss": 0.6329,
+      "step": 1001
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2513135693111399e-05,
+      "loss": 0.6269,
+      "step": 1002
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249947297237298e-05,
+      "loss": 0.6132,
+      "step": 1003
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485805273142737e-05,
+      "loss": 0.6406,
+      "step": 1004
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2472132622644222e-05,
+      "loss": 0.6258,
+      "step": 1005
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2458455048110844e-05,
+      "loss": 0.6304,
+      "step": 1006
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2444772576785828e-05,
+      "loss": 0.6399,
+      "step": 1007
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2431085235922143e-05,
+      "loss": 0.637,
+      "step": 1008
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.241739305278247e-05,
+      "loss": 0.6383,
+      "step": 1009
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2403696054639114e-05,
+      "loss": 0.6214,
+      "step": 1010
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2389994268773995e-05,
+      "loss": 0.6236,
+      "step": 1011
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376287722478553e-05,
+      "loss": 0.6354,
+      "step": 1012
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2362576443053716e-05,
+      "loss": 0.6369,
+      "step": 1013
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.234886045780984e-05,
+      "loss": 0.6249,
+      "step": 1014
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2335139794066645e-05,
+      "loss": 0.6259,
+      "step": 1015
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2321414479153191e-05,
+      "loss": 0.639,
+      "step": 1016
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2307684540407775e-05,
+      "loss": 0.633,
+      "step": 1017
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293950005177922e-05,
+      "loss": 0.6321,
+      "step": 1018
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280210900820309e-05,
+      "loss": 0.6194,
+      "step": 1019
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2266467254700712e-05,
+      "loss": 0.6233,
+      "step": 1020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225271909419395e-05,
+      "loss": 0.6495,
+      "step": 1021
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2238966446683846e-05,
+      "loss": 0.6249,
+      "step": 1022
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.6378,
+      "step": 1023
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2211447800233484e-05,
+      "loss": 0.6415,
+      "step": 1024
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2197681856105326e-05,
+      "loss": 0.6354,
+      "step": 1025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183911534597908e-05,
+      "loss": 0.6397,
+      "step": 1026
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2170136863139183e-05,
+      "loss": 0.6224,
+      "step": 1027
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156357869165771e-05,
+      "loss": 0.627,
+      "step": 1028
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2142574580122903e-05,
+      "loss": 0.626,
+      "step": 1029
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2128787023464364e-05,
+      "loss": 0.6239,
+      "step": 1030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114995226652437e-05,
+      "loss": 0.6383,
+      "step": 1031
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210119921715785e-05,
+      "loss": 0.6265,
+      "step": 1032
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2087399022459729e-05,
+      "loss": 0.6257,
+      "step": 1033
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073594670045526e-05,
+      "loss": 0.6253,
+      "step": 1034
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059786187410984e-05,
+      "loss": 0.6231,
+      "step": 1035
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2045973602060067e-05,
+      "loss": 0.6377,
+      "step": 1036
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2032156941504913e-05,
+      "loss": 0.6434,
+      "step": 1037
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018336233265775e-05,
+      "loss": 0.6249,
+      "step": 1038
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004511504870966e-05,
+      "loss": 0.6167,
+      "step": 1039
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1990682783856818e-05,
+      "loss": 0.6263,
+      "step": 1040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1976850097767598e-05,
+      "loss": 0.6414,
+      "step": 1041
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1963013474155487e-05,
+      "loss": 0.64,
+      "step": 1042
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1949172940580498e-05,
+      "loss": 0.6191,
+      "step": 1043
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1935328524610443e-05,
+      "loss": 0.6295,
+      "step": 1044
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1921480253820852e-05,
+      "loss": 0.6305,
+      "step": 1045
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1907628155794944e-05,
+      "loss": 0.6282,
+      "step": 1046
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1893772258123554e-05,
+      "loss": 0.6283,
+      "step": 1047
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1879912588405098e-05,
+      "loss": 0.6219,
+      "step": 1048
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186604917424549e-05,
+      "loss": 0.6367,
+      "step": 1049
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852182043258112e-05,
+      "loss": 0.6332,
+      "step": 1050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838311223063745e-05,
+      "loss": 0.6428,
+      "step": 1051
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1824436741290516e-05,
+      "loss": 0.6257,
+      "step": 1052
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1810558625573856e-05,
+      "loss": 0.6339,
+      "step": 1053
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1796676903556419e-05,
+      "loss": 0.6398,
+      "step": 1054
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1782791602888052e-05,
+      "loss": 0.6253,
+      "step": 1055
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176890275122573e-05,
+      "loss": 0.6329,
+      "step": 1056
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1755010376233498e-05,
+      "loss": 0.6396,
+      "step": 1057
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1741114505582418e-05,
+      "loss": 0.6444,
+      "step": 1058
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1727215166950519e-05,
+      "loss": 0.6345,
+      "step": 1059
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.171331238802273e-05,
+      "loss": 0.6404,
+      "step": 1060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169940619649084e-05,
+      "loss": 0.6219,
+      "step": 1061
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1685496620053434e-05,
+      "loss": 0.6345,
+      "step": 1062
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671583686415833e-05,
+      "loss": 0.6273,
+      "step": 1063
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657667423290055e-05,
+      "loss": 0.6285,
+      "step": 1064
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643747858394743e-05,
+      "loss": 0.6403,
+      "step": 1065
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629825019455117e-05,
+      "loss": 0.6293,
+      "step": 1066
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615898934202917e-05,
+      "loss": 0.6309,
+      "step": 1067
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1601969630376353e-05,
+      "loss": 0.63,
+      "step": 1068
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588037135720043e-05,
+      "loss": 0.6438,
+      "step": 1069
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1574101477984966e-05,
+      "loss": 0.6262,
+      "step": 1070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156016268492839e-05,
+      "loss": 0.6183,
+      "step": 1071
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1546220784313838e-05,
+      "loss": 0.624,
+      "step": 1072
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1532275803911021e-05,
+      "loss": 0.639,
+      "step": 1073
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151832777149578e-05,
+      "loss": 0.6322,
+      "step": 1074
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1504376714850041e-05,
+      "loss": 0.6261,
+      "step": 1075
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1490422661761744e-05,
+      "loss": 0.6383,
+      "step": 1076
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476465640024814e-05,
+      "loss": 0.6368,
+      "step": 1077
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1462505677439068e-05,
+      "loss": 0.6284,
+      "step": 1078
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1448542801810203e-05,
+      "loss": 0.6331,
+      "step": 1079
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434577040949699e-05,
+      "loss": 0.62,
+      "step": 1080
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1420608422674793e-05,
+      "loss": 0.6403,
+      "step": 1081
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1406636974808411e-05,
+      "loss": 0.6228,
+      "step": 1082
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1392662725179114e-05,
+      "loss": 0.6262,
+      "step": 1083
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1378685701621047e-05,
+      "loss": 0.6269,
+      "step": 1084
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1364705931973872e-05,
+      "loss": 0.6373,
+      "step": 1085
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1350723444082736e-05,
+      "loss": 0.6343,
+      "step": 1086
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1336738265798187e-05,
+      "loss": 0.635,
+      "step": 1087
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1322750424976133e-05,
+      "loss": 0.6196,
+      "step": 1088
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1308759949477786e-05,
+      "loss": 0.6301,
+      "step": 1089
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294766867169615e-05,
+      "loss": 0.6213,
+      "step": 1090
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1280771205923269e-05,
+      "loss": 0.6307,
+      "step": 1091
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1266772993615543e-05,
+      "loss": 0.6294,
+      "step": 1092
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1252772258128303e-05,
+      "loss": 0.6516,
+      "step": 1093
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238769027348453e-05,
+      "loss": 0.6488,
+      "step": 1094
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224763329167859e-05,
+      "loss": 0.6341,
+      "step": 1095
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1210755191483295e-05,
+      "loss": 0.6219,
+      "step": 1096
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1196744642196417e-05,
+      "loss": 0.6331,
+      "step": 1097
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1182731709213658e-05,
+      "loss": 0.637,
+      "step": 1098
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1168716420446219e-05,
+      "loss": 0.6365,
+      "step": 1099
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115469880380998e-05,
+      "loss": 0.6361,
+      "step": 1100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1140678887225468e-05,
+      "loss": 0.6262,
+      "step": 1101
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.112665669861778e-05,
+      "loss": 0.635,
+      "step": 1102
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1112632265916548e-05,
+      "loss": 0.6308,
+      "step": 1103
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1098605617055871e-05,
+      "loss": 0.6156,
+      "step": 1104
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1084576779974257e-05,
+      "loss": 0.6285,
+      "step": 1105
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.107054578261458e-05,
+      "loss": 0.628,
+      "step": 1106
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1056512652924014e-05,
+      "loss": 0.6367,
+      "step": 1107
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1042477418853983e-05,
+      "loss": 0.6391,
+      "step": 1108
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1028440108360092e-05,
+      "loss": 0.6285,
+      "step": 1109
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1014400749402096e-05,
+      "loss": 0.6234,
+      "step": 1110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1000359369943818e-05,
+      "loss": 0.6261,
+      "step": 1111
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986315997953118e-05,
+      "loss": 0.6216,
+      "step": 1112
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0972270661401812e-05,
+      "loss": 0.6359,
+      "step": 1113
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0958223388265638e-05,
+      "loss": 0.628,
+      "step": 1114
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.094417420652419e-05,
+      "loss": 0.6248,
+      "step": 1115
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930123144160857e-05,
+      "loss": 0.632,
+      "step": 1116
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091607022916278e-05,
+      "loss": 0.6294,
+      "step": 1117
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090201548952079e-05,
+      "loss": 0.6215,
+      "step": 1118
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0887958953229349e-05,
+      "loss": 0.6435,
+      "step": 1119
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0873900648286501e-05,
+      "loss": 0.6311,
+      "step": 1120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0859840602693813e-05,
+      "loss": 0.6429,
+      "step": 1121
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845778844456319e-05,
+      "loss": 0.6283,
+      "step": 1122
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0831715401582458e-05,
+      "loss": 0.6325,
+      "step": 1123
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081765030208403e-05,
+      "loss": 0.6349,
+      "step": 1124
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803583573976137e-05,
+      "loss": 0.6208,
+      "step": 1125
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.078951524527712e-05,
+      "loss": 0.6329,
+      "step": 1126
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077544534400851e-05,
+      "loss": 0.6319,
+      "step": 1127
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761373898194972e-05,
+      "loss": 0.6252,
+      "step": 1128
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.6205,
+      "step": 1129
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0733226485047089e-05,
+      "loss": 0.6174,
+      "step": 1130
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0719150573777226e-05,
+      "loss": 0.6223,
+      "step": 1131
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0705073230091296e-05,
+      "loss": 0.6197,
+      "step": 1132
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069099448202878e-05,
+      "loss": 0.6367,
+      "step": 1133
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0676914357631968e-05,
+      "loss": 0.6383,
+      "step": 1134
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0662832884945884e-05,
+      "loss": 0.6253,
+      "step": 1135
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0648750092018241e-05,
+      "loss": 0.6299,
+      "step": 1136
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0634666006899375e-05,
+      "loss": 0.6157,
+      "step": 1137
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0620580657642206e-05,
+      "loss": 0.6191,
+      "step": 1138
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0606494072302164e-05,
+      "loss": 0.6296,
+      "step": 1139
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0592406278937143e-05,
+      "loss": 0.6228,
+      "step": 1140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0578317305607451e-05,
+      "loss": 0.6339,
+      "step": 1141
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0564227180375727e-05,
+      "loss": 0.6283,
+      "step": 1142
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055013593130693e-05,
+      "loss": 0.6152,
+      "step": 1143
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0536043586468236e-05,
+      "loss": 0.637,
+      "step": 1144
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0521950173929017e-05,
+      "loss": 0.6208,
+      "step": 1145
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0507855721760757e-05,
+      "loss": 0.625,
+      "step": 1146
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049376025803703e-05,
+      "loss": 0.6313,
+      "step": 1147
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479663810833413e-05,
+      "loss": 0.6397,
+      "step": 1148
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046556640822744e-05,
+      "loss": 0.6233,
+      "step": 1149
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0451468078298554e-05,
+      "loss": 0.6299,
+      "step": 1150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0437368849128046e-05,
+      "loss": 0.6327,
+      "step": 1151
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0423268748798992e-05,
+      "loss": 0.6283,
+      "step": 1152
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0409167805396202e-05,
+      "loss": 0.6264,
+      "step": 1153
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039506604700618e-05,
+      "loss": 0.6307,
+      "step": 1154
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380963501717034e-05,
+      "loss": 0.6293,
+      "step": 1155
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366860197618451e-05,
+      "loss": 0.6202,
+      "step": 1156
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0352756162801626e-05,
+      "loss": 0.6243,
+      "step": 1157
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338651425359217e-05,
+      "loss": 0.6273,
+      "step": 1158
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0324546013385258e-05,
+      "loss": 0.6241,
+      "step": 1159
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0310439954975162e-05,
+      "loss": 0.6253,
+      "step": 1160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0296333278225599e-05,
+      "loss": 0.6177,
+      "step": 1161
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0282226011234483e-05,
+      "loss": 0.63,
+      "step": 1162
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026811818210091e-05,
+      "loss": 0.628,
+      "step": 1163
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254009818925082e-05,
+      "loss": 0.6266,
+      "step": 1164
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0239900949808274e-05,
+      "loss": 0.6346,
+      "step": 1165
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0225791602852764e-05,
+      "loss": 0.6248,
+      "step": 1166
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0211681806161787e-05,
+      "loss": 0.6255,
+      "step": 1167
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0197571587839466e-05,
+      "loss": 0.6274,
+      "step": 1168
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0183460975990773e-05,
+      "loss": 0.6321,
+      "step": 1169
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0169349998721459e-05,
+      "loss": 0.6266,
+      "step": 1170
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0155238684138e-05,
+      "loss": 0.63,
+      "step": 1171
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0141127060347549e-05,
+      "loss": 0.6351,
+      "step": 1172
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127015155457875e-05,
+      "loss": 0.6305,
+      "step": 1173
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01129029975773e-05,
+      "loss": 0.6135,
+      "step": 1174
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0098790614814658e-05,
+      "loss": 0.6376,
+      "step": 1175
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0084678035279227e-05,
+      "loss": 0.6231,
+      "step": 1176
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0070565287080676e-05,
+      "loss": 0.6321,
+      "step": 1177
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0056452398329015e-05,
+      "loss": 0.6168,
+      "step": 1178
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0042339397134528e-05,
+      "loss": 0.6232,
+      "step": 1179
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0028226311607727e-05,
+      "loss": 0.6361,
+      "step": 1180
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0014113169859285e-05,
+      "loss": 0.6377,
+      "step": 1181
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.6217,
+      "step": 1182
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.985886830140717e-06,
+      "loss": 0.6285,
+      "step": 1183
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.971773688392275e-06,
+      "loss": 0.6293,
+      "step": 1184
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.957660602865477e-06,
+      "loss": 0.6323,
+      "step": 1185
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.943547601670985e-06,
+      "loss": 0.6279,
+      "step": 1186
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.929434712919327e-06,
+      "loss": 0.6289,
+      "step": 1187
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.915321964720773e-06,
+      "loss": 0.6435,
+      "step": 1188
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901209385185345e-06,
+      "loss": 0.6349,
+      "step": 1189
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.887097002422705e-06,
+      "loss": 0.631,
+      "step": 1190
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872984844542128e-06,
+      "loss": 0.6298,
+      "step": 1191
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.858872939652452e-06,
+      "loss": 0.6271,
+      "step": 1192
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.844761315862002e-06,
+      "loss": 0.6339,
+      "step": 1193
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.830650001278543e-06,
+      "loss": 0.6236,
+      "step": 1194
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.816539024009227e-06,
+      "loss": 0.6366,
+      "step": 1195
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.802428412160538e-06,
+      "loss": 0.6174,
+      "step": 1196
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788318193838218e-06,
+      "loss": 0.6345,
+      "step": 1197
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.774208397147239e-06,
+      "loss": 0.6286,
+      "step": 1198
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76009905019173e-06,
+      "loss": 0.6292,
+      "step": 1199
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74599018107492e-06,
+      "loss": 0.6232,
+      "step": 1200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.731881817899092e-06,
+      "loss": 0.6319,
+      "step": 1201
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717773988765517e-06,
+      "loss": 0.6359,
+      "step": 1202
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.703666721774403e-06,
+      "loss": 0.6161,
+      "step": 1203
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.689560045024843e-06,
+      "loss": 0.6427,
+      "step": 1204
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.675453986614743e-06,
+      "loss": 0.6335,
+      "step": 1205
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.66134857464079e-06,
+      "loss": 0.636,
+      "step": 1206
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647243837198375e-06,
+      "loss": 0.6338,
+      "step": 1207
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.633139802381552e-06,
+      "loss": 0.6416,
+      "step": 1208
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619036498282968e-06,
+      "loss": 0.6251,
+      "step": 1209
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.604933952993822e-06,
+      "loss": 0.6123,
+      "step": 1210
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.590832194603801e-06,
+      "loss": 0.6383,
+      "step": 1211
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57673125120101e-06,
+      "loss": 0.6397,
+      "step": 1212
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.562631150871959e-06,
+      "loss": 0.6227,
+      "step": 1213
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.548531921701448e-06,
+      "loss": 0.6307,
+      "step": 1214
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534433591772562e-06,
+      "loss": 0.629,
+      "step": 1215
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.520336189166589e-06,
+      "loss": 0.6121,
+      "step": 1216
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.506239741962971e-06,
+      "loss": 0.6279,
+      "step": 1217
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.492144278239244e-06,
+      "loss": 0.6257,
+      "step": 1218
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.478049826070988e-06,
+      "loss": 0.6181,
+      "step": 1219
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.463956413531769e-06,
+      "loss": 0.6171,
+      "step": 1220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.449864068693072e-06,
+      "loss": 0.6152,
+      "step": 1221
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.435772819624274e-06,
+      "loss": 0.6294,
+      "step": 1222
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42168269439255e-06,
+      "loss": 0.6248,
+      "step": 1223
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.407593721062858e-06,
+      "loss": 0.6262,
+      "step": 1224
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39350592769784e-06,
+      "loss": 0.6366,
+      "step": 1225
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.379419342357797e-06,
+      "loss": 0.6264,
+      "step": 1226
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.365333993100628e-06,
+      "loss": 0.6373,
+      "step": 1227
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.351249907981762e-06,
+      "loss": 0.6246,
+      "step": 1228
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.33716711505412e-06,
+      "loss": 0.6078,
+      "step": 1229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.323085642368032e-06,
+      "loss": 0.619,
+      "step": 1230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309005517971222e-06,
+      "loss": 0.6286,
+      "step": 1231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29492676990871e-06,
+      "loss": 0.618,
+      "step": 1232
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280849426222778e-06,
+      "loss": 0.6157,
+      "step": 1233
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266773514952917e-06,
+      "loss": 0.6218,
+      "step": 1234
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 0.6133,
+      "step": 1235
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.23862610180503e-06,
+      "loss": 0.6153,
+      "step": 1236
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.224554655991492e-06,
+      "loss": 0.6234,
+      "step": 1237
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210484754722882e-06,
+      "loss": 0.6289,
+      "step": 1238
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196416426023868e-06,
+      "loss": 0.6241,
+      "step": 1239
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182349697915972e-06,
+      "loss": 0.6342,
+      "step": 1240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.168284598417547e-06,
+      "loss": 0.6212,
+      "step": 1241
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154221155543684e-06,
+      "loss": 0.6181,
+      "step": 1242
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.140159397306188e-06,
+      "loss": 0.6242,
+      "step": 1243
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.126099351713499e-06,
+      "loss": 0.6219,
+      "step": 1244
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112041046770653e-06,
+      "loss": 0.6213,
+      "step": 1245
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.097984510479216e-06,
+      "loss": 0.6104,
+      "step": 1246
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.083929770837222e-06,
+      "loss": 0.6253,
+      "step": 1247
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.069876855839148e-06,
+      "loss": 0.6296,
+      "step": 1248
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055825793475814e-06,
+      "loss": 0.6179,
+      "step": 1249
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.041776611734364e-06,
+      "loss": 0.6324,
+      "step": 1250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.027729338598188e-06,
+      "loss": 0.6317,
+      "step": 1251
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.013684002046883e-06,
+      "loss": 0.6403,
+      "step": 1252
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.999640630056183e-06,
+      "loss": 0.6311,
+      "step": 1253
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.985599250597906e-06,
+      "loss": 0.6294,
+      "step": 1254
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.971559891639913e-06,
+      "loss": 0.6196,
+      "step": 1255
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.957522581146019e-06,
+      "loss": 0.6286,
+      "step": 1256
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.943487347075988e-06,
+      "loss": 0.6265,
+      "step": 1257
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.929454217385423e-06,
+      "loss": 0.6247,
+      "step": 1258
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.915423220025747e-06,
+      "loss": 0.6192,
+      "step": 1259
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.901394382944135e-06,
+      "loss": 0.6035,
+      "step": 1260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887367734083454e-06,
+      "loss": 0.6239,
+      "step": 1261
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.873343301382223e-06,
+      "loss": 0.6225,
+      "step": 1262
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.859321112774535e-06,
+      "loss": 0.6289,
+      "step": 1263
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.845301196190023e-06,
+      "loss": 0.6174,
+      "step": 1264
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.831283579553781e-06,
+      "loss": 0.6366,
+      "step": 1265
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.817268290786343e-06,
+      "loss": 0.6253,
+      "step": 1266
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.803255357803584e-06,
+      "loss": 0.6325,
+      "step": 1267
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789244808516707e-06,
+      "loss": 0.62,
+      "step": 1268
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.775236670832146e-06,
+      "loss": 0.6246,
+      "step": 1269
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761230972651549e-06,
+      "loss": 0.6271,
+      "step": 1270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.747227741871698e-06,
+      "loss": 0.6181,
+      "step": 1271
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.733227006384459e-06,
+      "loss": 0.6281,
+      "step": 1272
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719228794076733e-06,
+      "loss": 0.6211,
+      "step": 1273
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.705233132830383e-06,
+      "loss": 0.6306,
+      "step": 1274
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.691240050522215e-06,
+      "loss": 0.6276,
+      "step": 1275
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.677249575023874e-06,
+      "loss": 0.6281,
+      "step": 1276
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.663261734201818e-06,
+      "loss": 0.6253,
+      "step": 1277
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.649276555917267e-06,
+      "loss": 0.6259,
+      "step": 1278
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63529406802613e-06,
+      "loss": 0.6278,
+      "step": 1279
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.621314298378958e-06,
+      "loss": 0.628,
+      "step": 1280
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607337274820888e-06,
+      "loss": 0.626,
+      "step": 1281
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.593363025191592e-06,
+      "loss": 0.6305,
+      "step": 1282
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.57939157732521e-06,
+      "loss": 0.623,
+      "step": 1283
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565422959050303e-06,
+      "loss": 0.6282,
+      "step": 1284
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.551457198189799e-06,
+      "loss": 0.6314,
+      "step": 1285
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53749432256093e-06,
+      "loss": 0.6221,
+      "step": 1286
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52353435997519e-06,
+      "loss": 0.6258,
+      "step": 1287
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.509577338238255e-06,
+      "loss": 0.6145,
+      "step": 1288
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.495623285149962e-06,
+      "loss": 0.6236,
+      "step": 1289
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.481672228504224e-06,
+      "loss": 0.6424,
+      "step": 1290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46772419608898e-06,
+      "loss": 0.6158,
+      "step": 1291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453779215686165e-06,
+      "loss": 0.6239,
+      "step": 1292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.439837315071612e-06,
+      "loss": 0.6204,
+      "step": 1293
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.425898522015038e-06,
+      "loss": 0.6372,
+      "step": 1294
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.411962864279957e-06,
+      "loss": 0.6212,
+      "step": 1295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39803036962365e-06,
+      "loss": 0.6212,
+      "step": 1296
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.384101065797087e-06,
+      "loss": 0.6242,
+      "step": 1297
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.370174980544885e-06,
+      "loss": 0.6209,
+      "step": 1298
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35625214160526e-06,
+      "loss": 0.6244,
+      "step": 1299
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.342332576709943e-06,
+      "loss": 0.6274,
+      "step": 1300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328416313584169e-06,
+      "loss": 0.6325,
+      "step": 1301
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.314503379946569e-06,
+      "loss": 0.6276,
+      "step": 1302
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.300593803509163e-06,
+      "loss": 0.628,
+      "step": 1303
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.286687611977275e-06,
+      "loss": 0.6184,
+      "step": 1304
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.272784833049485e-06,
+      "loss": 0.6271,
+      "step": 1305
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.258885494417584e-06,
+      "loss": 0.6392,
+      "step": 1306
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.244989623766502e-06,
+      "loss": 0.6179,
+      "step": 1307
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.231097248774273e-06,
+      "loss": 0.6274,
+      "step": 1308
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.217208397111948e-06,
+      "loss": 0.6139,
+      "step": 1309
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.203323096443586e-06,
+      "loss": 0.6163,
+      "step": 1310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18944137442615e-06,
+      "loss": 0.6455,
+      "step": 1311
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.175563258709485e-06,
+      "loss": 0.6252,
+      "step": 1312
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.161688776936259e-06,
+      "loss": 0.6242,
+      "step": 1313
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14781795674189e-06,
+      "loss": 0.6235,
+      "step": 1314
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.133950825754511e-06,
+      "loss": 0.628,
+      "step": 1315
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.120087411594903e-06,
+      "loss": 0.612,
+      "step": 1316
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.106227741876447e-06,
+      "loss": 0.6142,
+      "step": 1317
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.092371844205063e-06,
+      "loss": 0.6199,
+      "step": 1318
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.078519746179153e-06,
+      "loss": 0.6254,
+      "step": 1319
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.064671475389562e-06,
+      "loss": 0.6181,
+      "step": 1320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.050827059419502e-06,
+      "loss": 0.6058,
+      "step": 1321
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.036986525844516e-06,
+      "loss": 0.6151,
+      "step": 1322
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.023149902232404e-06,
+      "loss": 0.6256,
+      "step": 1323
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.009317216143187e-06,
+      "loss": 0.6193,
+      "step": 1324
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995488495129039e-06,
+      "loss": 0.6187,
+      "step": 1325
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.981663766734228e-06,
+      "loss": 0.6192,
+      "step": 1326
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.967843058495092e-06,
+      "loss": 0.6143,
+      "step": 1327
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.954026397939934e-06,
+      "loss": 0.6201,
+      "step": 1328
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940213812589018e-06,
+      "loss": 0.6072,
+      "step": 1329
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.926405329954474e-06,
+      "loss": 0.6254,
+      "step": 1330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.912600977540275e-06,
+      "loss": 0.6109,
+      "step": 1331
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.898800782842153e-06,
+      "loss": 0.6345,
+      "step": 1332
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885004773347565e-06,
+      "loss": 0.6287,
+      "step": 1333
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87121297653564e-06,
+      "loss": 0.6238,
+      "step": 1334
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.857425419877097e-06,
+      "loss": 0.6297,
+      "step": 1335
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.843642130834232e-06,
+      "loss": 0.6179,
+      "step": 1336
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82986313686082e-06,
+      "loss": 0.6224,
+      "step": 1337
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.816088465402097e-06,
+      "loss": 0.6246,
+      "step": 1338
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802318143894678e-06,
+      "loss": 0.632,
+      "step": 1339
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.788552199766518e-06,
+      "loss": 0.6173,
+      "step": 1340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.6139,
+      "step": 1341
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.761033553316156e-06,
+      "loss": 0.6279,
+      "step": 1342
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.747280905806051e-06,
+      "loss": 0.6239,
+      "step": 1343
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.73353274529929e-06,
+      "loss": 0.6211,
+      "step": 1344
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719789099179696e-06,
+      "loss": 0.6239,
+      "step": 1345
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.706049994822083e-06,
+      "loss": 0.6176,
+      "step": 1346
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69231545959223e-06,
+      "loss": 0.6229,
+      "step": 1347
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.678585520846816e-06,
+      "loss": 0.6224,
+      "step": 1348
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.664860205933356e-06,
+      "loss": 0.623,
+      "step": 1349
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.651139542190164e-06,
+      "loss": 0.634,
+      "step": 1350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.637423556946284e-06,
+      "loss": 0.6275,
+      "step": 1351
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.623712277521451e-06,
+      "loss": 0.6204,
+      "step": 1352
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.610005731226009e-06,
+      "loss": 0.6136,
+      "step": 1353
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5963039453608875e-06,
+      "loss": 0.6259,
+      "step": 1354
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.582606947217537e-06,
+      "loss": 0.6232,
+      "step": 1355
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.568914764077857e-06,
+      "loss": 0.6275,
+      "step": 1356
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555227423214174e-06,
+      "loss": 0.6194,
+      "step": 1357
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.541544951889156e-06,
+      "loss": 0.6191,
+      "step": 1358
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.52786737735578e-06,
+      "loss": 0.6195,
+      "step": 1359
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514194726857268e-06,
+      "loss": 0.623,
+      "step": 1360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500527027627025e-06,
+      "loss": 0.6206,
+      "step": 1361
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.486864306888608e-06,
+      "loss": 0.6197,
+      "step": 1362
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.473206591855646e-06,
+      "loss": 0.6157,
+      "step": 1363
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.45955390973181e-06,
+      "loss": 0.6254,
+      "step": 1364
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.445906287710733e-06,
+      "loss": 0.6314,
+      "step": 1365
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.432263752975981e-06,
+      "loss": 0.6215,
+      "step": 1366
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.41862633270098e-06,
+      "loss": 0.619,
+      "step": 1367
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.404994054048962e-06,
+      "loss": 0.6217,
+      "step": 1368
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.391366944172941e-06,
+      "loss": 0.6172,
+      "step": 1369
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.377745030215606e-06,
+      "loss": 0.6173,
+      "step": 1370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.364128339309326e-06,
+      "loss": 0.6286,
+      "step": 1371
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.350516898576041e-06,
+      "loss": 0.627,
+      "step": 1372
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.33691073512725e-06,
+      "loss": 0.6136,
+      "step": 1373
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.323309876063936e-06,
+      "loss": 0.6222,
+      "step": 1374
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309714348476513e-06,
+      "loss": 0.6205,
+      "step": 1375
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.296124179444784e-06,
+      "loss": 0.625,
+      "step": 1376
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.282539396037868e-06,
+      "loss": 0.6266,
+      "step": 1377
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.268960025314169e-06,
+      "loss": 0.6214,
+      "step": 1378
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.255386094321293e-06,
+      "loss": 0.6218,
+      "step": 1379
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.241817630096035e-06,
+      "loss": 0.628,
+      "step": 1380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.228254659664278e-06,
+      "loss": 0.6182,
+      "step": 1381
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.214697210040973e-06,
+      "loss": 0.6324,
+      "step": 1382
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.201145308230075e-06,
+      "loss": 0.6443,
+      "step": 1383
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.187598981224488e-06,
+      "loss": 0.6185,
+      "step": 1384
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.174058256006012e-06,
+      "loss": 0.6119,
+      "step": 1385
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.160523159545285e-06,
+      "loss": 0.6242,
+      "step": 1386
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1469937188017444e-06,
+      "loss": 0.6243,
+      "step": 1387
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133469960723549e-06,
+      "loss": 0.6136,
+      "step": 1388
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.119951912247545e-06,
+      "loss": 0.6305,
+      "step": 1389
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1064396002992075e-06,
+      "loss": 0.6094,
+      "step": 1390
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.092933051792583e-06,
+      "loss": 0.6294,
+      "step": 1391
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079432293630244e-06,
+      "loss": 0.6294,
+      "step": 1392
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065937352703218e-06,
+      "loss": 0.6062,
+      "step": 1393
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.052448255890958e-06,
+      "loss": 0.622,
+      "step": 1394
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.038965030061273e-06,
+      "loss": 0.6249,
+      "step": 1395
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0254877020702686e-06,
+      "loss": 0.6229,
+      "step": 1396
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.012016298762317e-06,
+      "loss": 0.6231,
+      "step": 1397
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.998550846969981e-06,
+      "loss": 0.6208,
+      "step": 1398
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.985091373513972e-06,
+      "loss": 0.6193,
+      "step": 1399
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9716379052030914e-06,
+      "loss": 0.6225,
+      "step": 1400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9581904688341854e-06,
+      "loss": 0.6191,
+      "step": 1401
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.94474909119208e-06,
+      "loss": 0.6214,
+      "step": 1402
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.931313799049526e-06,
+      "loss": 0.6203,
+      "step": 1403
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.917884619167177e-06,
+      "loss": 0.6254,
+      "step": 1404
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.904461578293483e-06,
+      "loss": 0.6188,
+      "step": 1405
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8910447031646884e-06,
+      "loss": 0.6218,
+      "step": 1406
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8776340205047446e-06,
+      "loss": 0.6224,
+      "step": 1407
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.864229557025273e-06,
+      "loss": 0.6234,
+      "step": 1408
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.850831339425508e-06,
+      "loss": 0.6267,
+      "step": 1409
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837439394392238e-06,
+      "loss": 0.6189,
+      "step": 1410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8240537485997704e-06,
+      "loss": 0.6172,
+      "step": 1411
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.810674428709845e-06,
+      "loss": 0.6227,
+      "step": 1412
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797301461371626e-06,
+      "loss": 0.6201,
+      "step": 1413
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7839348732216e-06,
+      "loss": 0.6302,
+      "step": 1414
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7705746908835734e-06,
+      "loss": 0.6219,
+      "step": 1415
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.757220940968567e-06,
+      "loss": 0.6153,
+      "step": 1416
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.743873650074807e-06,
+      "loss": 0.6137,
+      "step": 1417
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.73053284478765e-06,
+      "loss": 0.6106,
+      "step": 1418
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7171985516795315e-06,
+      "loss": 0.6205,
+      "step": 1419
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.703870797309922e-06,
+      "loss": 0.6186,
+      "step": 1420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.690549608225258e-06,
+      "loss": 0.6228,
+      "step": 1421
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677235010958916e-06,
+      "loss": 0.6183,
+      "step": 1422
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.663927032031118e-06,
+      "loss": 0.6155,
+      "step": 1423
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.650625697948932e-06,
+      "loss": 0.6249,
+      "step": 1424
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.637331035206166e-06,
+      "loss": 0.6223,
+      "step": 1425
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.624043070283352e-06,
+      "loss": 0.6134,
+      "step": 1426
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.610761829647685e-06,
+      "loss": 0.6176,
+      "step": 1427
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.597487339752953e-06,
+      "loss": 0.635,
+      "step": 1428
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.584219627039513e-06,
+      "loss": 0.6267,
+      "step": 1429
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.570958717934211e-06,
+      "loss": 0.6216,
+      "step": 1430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.557704638850352e-06,
+      "loss": 0.6165,
+      "step": 1431
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.544457416187624e-06,
+      "loss": 0.6058,
+      "step": 1432
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531217076332068e-06,
+      "loss": 0.6118,
+      "step": 1433
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.517983645656014e-06,
+      "loss": 0.6234,
+      "step": 1434
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5047571505180265e-06,
+      "loss": 0.6137,
+      "step": 1435
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.491537617262864e-06,
+      "loss": 0.6195,
+      "step": 1436
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4783250722214066e-06,
+      "loss": 0.6184,
+      "step": 1437
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.465119541710625e-06,
+      "loss": 0.6235,
+      "step": 1438
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451921052033516e-06,
+      "loss": 0.6056,
+      "step": 1439
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438729629479045e-06,
+      "loss": 0.6216,
+      "step": 1440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4255453003221115e-06,
+      "loss": 0.6381,
+      "step": 1441
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.412368090823477e-06,
+      "loss": 0.6176,
+      "step": 1442
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.399198027229732e-06,
+      "loss": 0.6247,
+      "step": 1443
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.386035135773222e-06,
+      "loss": 0.6254,
+      "step": 1444
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37287944267202e-06,
+      "loss": 0.6221,
+      "step": 1445
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359730974129852e-06,
+      "loss": 0.615,
+      "step": 1446
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 0.6182,
+      "step": 1447
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33345581546552e-06,
+      "loss": 0.6172,
+      "step": 1448
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.320329177678656e-06,
+      "loss": 0.6248,
+      "step": 1449
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.307209869121318e-06,
+      "loss": 0.6319,
+      "step": 1450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.29409791592476e-06,
+      "loss": 0.6067,
+      "step": 1451
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28099334420559e-06,
+      "loss": 0.6203,
+      "step": 1452
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267896180065711e-06,
+      "loss": 0.6185,
+      "step": 1453
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.254806449592272e-06,
+      "loss": 0.6234,
+      "step": 1454
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241724178857621e-06,
+      "loss": 0.6232,
+      "step": 1455
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.228649393919231e-06,
+      "loss": 0.6073,
+      "step": 1456
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.21558212081969e-06,
+      "loss": 0.625,
+      "step": 1457
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.202522385586595e-06,
+      "loss": 0.6143,
+      "step": 1458
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.18947021423256e-06,
+      "loss": 0.622,
+      "step": 1459
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176425632755105e-06,
+      "loss": 0.6069,
+      "step": 1460
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.163388667136646e-06,
+      "loss": 0.6187,
+      "step": 1461
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1503593433444316e-06,
+      "loss": 0.6142,
+      "step": 1462
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1373376873304814e-06,
+      "loss": 0.6141,
+      "step": 1463
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12432372503155e-06,
+      "loss": 0.6067,
+      "step": 1464
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1113174823690615e-06,
+      "loss": 0.6235,
+      "step": 1465
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.0983189852490695e-06,
+      "loss": 0.6328,
+      "step": 1466
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.085328259562195e-06,
+      "loss": 0.6214,
+      "step": 1467
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.072345331183578e-06,
+      "loss": 0.6194,
+      "step": 1468
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.059370225972834e-06,
+      "loss": 0.6178,
+      "step": 1469
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.046402969773992e-06,
+      "loss": 0.6131,
+      "step": 1470
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0334435884154526e-06,
+      "loss": 0.6249,
+      "step": 1471
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0204921077099225e-06,
+      "loss": 0.6264,
+      "step": 1472
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.007548553454379e-06,
+      "loss": 0.617,
+      "step": 1473
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.99461295143001e-06,
+      "loss": 0.6311,
+      "step": 1474
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981685327402156e-06,
+      "loss": 0.6227,
+      "step": 1475
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96876570712028e-06,
+      "loss": 0.6199,
+      "step": 1476
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9558541163178915e-06,
+      "loss": 0.6222,
+      "step": 1477
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.942950580712519e-06,
+      "loss": 0.6216,
+      "step": 1478
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.93005512600563e-06,
+      "loss": 0.6231,
+      "step": 1479
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.917167777882613e-06,
+      "loss": 0.617,
+      "step": 1480
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.904288562012703e-06,
+      "loss": 0.6202,
+      "step": 1481
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.891417504048926e-06,
+      "loss": 0.6068,
+      "step": 1482
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.878554629628081e-06,
+      "loss": 0.6132,
+      "step": 1483
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.865699964370648e-06,
+      "loss": 0.6126,
+      "step": 1484
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852853533880768e-06,
+      "loss": 0.6371,
+      "step": 1485
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.840015363746169e-06,
+      "loss": 0.6212,
+      "step": 1486
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827185479538138e-06,
+      "loss": 0.624,
+      "step": 1487
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8143639068114465e-06,
+      "loss": 0.6133,
+      "step": 1488
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.801550671104319e-06,
+      "loss": 0.6187,
+      "step": 1489
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788745797938372e-06,
+      "loss": 0.6182,
+      "step": 1490
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7759493128185584e-06,
+      "loss": 0.6031,
+      "step": 1491
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7631612412331395e-06,
+      "loss": 0.6252,
+      "step": 1492
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.750381608653605e-06,
+      "loss": 0.6208,
+      "step": 1493
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7376104405346376e-06,
+      "loss": 0.6154,
+      "step": 1494
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7248477623140655e-06,
+      "loss": 0.6289,
+      "step": 1495
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.712093599412802e-06,
+      "loss": 0.6053,
+      "step": 1496
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.699347977234799e-06,
+      "loss": 0.6245,
+      "step": 1497
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.686610921166996e-06,
+      "loss": 0.6257,
+      "step": 1498
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.673882456579282e-06,
+      "loss": 0.6062,
+      "step": 1499
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.6166,
+      "step": 1500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648451403238013e-06,
+      "loss": 0.6134,
+      "step": 1501
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.635748865138453e-06,
+      "loss": 0.6153,
+      "step": 1502
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.623055019826862e-06,
+      "loss": 0.6186,
+      "step": 1503
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.610369892587064e-06,
+      "loss": 0.6111,
+      "step": 1504
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5976935086854914e-06,
+      "loss": 0.6126,
+      "step": 1505
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.585025893371187e-06,
+      "loss": 0.6216,
+      "step": 1506
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.572367071875715e-06,
+      "loss": 0.6149,
+      "step": 1507
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5597170694131245e-06,
+      "loss": 0.6172,
+      "step": 1508
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.547075911179902e-06,
+      "loss": 0.6158,
+      "step": 1509
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5344436223549135e-06,
+      "loss": 0.6142,
+      "step": 1510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5218202280993725e-06,
+      "loss": 0.6202,
+      "step": 1511
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.50920575355675e-06,
+      "loss": 0.6289,
+      "step": 1512
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49660022385278e-06,
+      "loss": 0.6161,
+      "step": 1513
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.484003664095354e-06,
+      "loss": 0.6141,
+      "step": 1514
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471416099374525e-06,
+      "loss": 0.6212,
+      "step": 1515
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4588375547624e-06,
+      "loss": 0.6198,
+      "step": 1516
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.446268055313132e-06,
+      "loss": 0.6195,
+      "step": 1517
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.43370762606287e-06,
+      "loss": 0.6236,
+      "step": 1518
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.42115629202968e-06,
+      "loss": 0.6194,
+      "step": 1519
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.408614078213519e-06,
+      "loss": 0.6184,
+      "step": 1520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3960810095961705e-06,
+      "loss": 0.6162,
+      "step": 1521
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.383557111141224e-06,
+      "loss": 0.6144,
+      "step": 1522
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.371042407793974e-06,
+      "loss": 0.6105,
+      "step": 1523
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.358536924481416e-06,
+      "loss": 0.6259,
+      "step": 1524
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.346040686112189e-06,
+      "loss": 0.6201,
+      "step": 1525
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333553717576503e-06,
+      "loss": 0.6255,
+      "step": 1526
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.321076043746108e-06,
+      "loss": 0.6118,
+      "step": 1527
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.308607689474241e-06,
+      "loss": 0.6139,
+      "step": 1528
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296148679595583e-06,
+      "loss": 0.6198,
+      "step": 1529
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283699038926199e-06,
+      "loss": 0.6151,
+      "step": 1530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.271258792263476e-06,
+      "loss": 0.6147,
+      "step": 1531
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25882796438612e-06,
+      "loss": 0.6025,
+      "step": 1532
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246406580054051e-06,
+      "loss": 0.5553,
+      "step": 1533
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2339946640083936e-06,
+      "loss": 0.5537,
+      "step": 1534
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.221592240971403e-06,
+      "loss": 0.5758,
+      "step": 1535
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.20919933564644e-06,
+      "loss": 0.5617,
+      "step": 1536
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.196815972717897e-06,
+      "loss": 0.5536,
+      "step": 1537
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.184442176851154e-06,
+      "loss": 0.5393,
+      "step": 1538
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.172077972692553e-06,
+      "loss": 0.5551,
+      "step": 1539
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1597233848693176e-06,
+      "loss": 0.5599,
+      "step": 1540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.147378437989522e-06,
+      "loss": 0.5776,
+      "step": 1541
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1350431566420326e-06,
+      "loss": 0.5571,
+      "step": 1542
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.122717565396474e-06,
+      "loss": 0.5776,
+      "step": 1543
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.110401688803164e-06,
+      "loss": 0.5566,
+      "step": 1544
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.098095551393066e-06,
+      "loss": 0.5627,
+      "step": 1545
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08579917767775e-06,
+      "loss": 0.5579,
+      "step": 1546
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.073512592149334e-06,
+      "loss": 0.559,
+      "step": 1547
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.061235819280456e-06,
+      "loss": 0.5498,
+      "step": 1548
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.048968883524182e-06,
+      "loss": 0.553,
+      "step": 1549
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.036711809314007e-06,
+      "loss": 0.5603,
+      "step": 1550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.024464621063773e-06,
+      "loss": 0.555,
+      "step": 1551
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.012227343167635e-06,
+      "loss": 0.5575,
+      "step": 1552
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.55,
+      "step": 1553
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9877826159154995e-06,
+      "loss": 0.5472,
+      "step": 1554
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.975575215248926e-06,
+      "loss": 0.5514,
+      "step": 1555
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.963377822315171e-06,
+      "loss": 0.5605,
+      "step": 1556
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.951190461409214e-06,
+      "loss": 0.5449,
+      "step": 1557
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.93901315680604e-06,
+      "loss": 0.5649,
+      "step": 1558
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.926845932760609e-06,
+      "loss": 0.5586,
+      "step": 1559
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.914688813507798e-06,
+      "loss": 0.5618,
+      "step": 1560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902541823262356e-06,
+      "loss": 0.5651,
+      "step": 1561
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.890404986218867e-06,
+      "loss": 0.5548,
+      "step": 1562
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878278326551682e-06,
+      "loss": 0.5459,
+      "step": 1563
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866161868414883e-06,
+      "loss": 0.5568,
+      "step": 1564
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8540556359422335e-06,
+      "loss": 0.5513,
+      "step": 1565
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8419596532471266e-06,
+      "loss": 0.5503,
+      "step": 1566
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.829873944422544e-06,
+      "loss": 0.5578,
+      "step": 1567
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.817798533540996e-06,
+      "loss": 0.5506,
+      "step": 1568
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805733444654496e-06,
+      "loss": 0.549,
+      "step": 1569
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.793678701794483e-06,
+      "loss": 0.5475,
+      "step": 1570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.781634328971796e-06,
+      "loss": 0.5545,
+      "step": 1571
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7696003501766155e-06,
+      "loss": 0.5557,
+      "step": 1572
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7575767893784174e-06,
+      "loss": 0.5459,
+      "step": 1573
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.745563670525942e-06,
+      "loss": 0.5564,
+      "step": 1574
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733561017547104e-06,
+      "loss": 0.5573,
+      "step": 1575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.721568854348998e-06,
+      "loss": 0.5639,
+      "step": 1576
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709587204817809e-06,
+      "loss": 0.5572,
+      "step": 1577
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697616092818789e-06,
+      "loss": 0.5482,
+      "step": 1578
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685655542196194e-06,
+      "loss": 0.5553,
+      "step": 1579
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.673705576773247e-06,
+      "loss": 0.5625,
+      "step": 1580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.661766220352098e-06,
+      "loss": 0.5773,
+      "step": 1581
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.649837496713739e-06,
+      "loss": 0.561,
+      "step": 1582
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.637919429618014e-06,
+      "loss": 0.5511,
+      "step": 1583
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.626012042803521e-06,
+      "loss": 0.5505,
+      "step": 1584
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.614115359987595e-06,
+      "loss": 0.5564,
+      "step": 1585
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.602229404866238e-06,
+      "loss": 0.5604,
+      "step": 1586
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.590354201114103e-06,
+      "loss": 0.5515,
+      "step": 1587
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.578489772384415e-06,
+      "loss": 0.5588,
+      "step": 1588
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.566636142308939e-06,
+      "loss": 0.5593,
+      "step": 1589
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.554793334497933e-06,
+      "loss": 0.5627,
+      "step": 1590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.542961372540096e-06,
+      "loss": 0.5611,
+      "step": 1591
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531140280002536e-06,
+      "loss": 0.5492,
+      "step": 1592
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.519330080430687e-06,
+      "loss": 0.55,
+      "step": 1593
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.507530797348313e-06,
+      "loss": 0.5473,
+      "step": 1594
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.495742454257418e-06,
+      "loss": 0.5549,
+      "step": 1595
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.483965074638219e-06,
+      "loss": 0.5462,
+      "step": 1596
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.472198681949098e-06,
+      "loss": 0.5567,
+      "step": 1597
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.460443299626547e-06,
+      "loss": 0.5544,
+      "step": 1598
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.448698951085143e-06,
+      "loss": 0.5532,
+      "step": 1599
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436965659717461e-06,
+      "loss": 0.558,
+      "step": 1600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.425243448894074e-06,
+      "loss": 0.559,
+      "step": 1601
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.413532341963477e-06,
+      "loss": 0.5434,
+      "step": 1602
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.401832362252044e-06,
+      "loss": 0.5538,
+      "step": 1603
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.390143533063991e-06,
+      "loss": 0.5624,
+      "step": 1604
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.378465877681317e-06,
+      "loss": 0.5444,
+      "step": 1605
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3667994193637794e-06,
+      "loss": 0.549,
+      "step": 1606
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.355144181348819e-06,
+      "loss": 0.5432,
+      "step": 1607
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.343500186851534e-06,
+      "loss": 0.5533,
+      "step": 1608
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331867459064623e-06,
+      "loss": 0.5635,
+      "step": 1609
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.320246021158351e-06,
+      "loss": 0.5478,
+      "step": 1610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3086358962804885e-06,
+      "loss": 0.5481,
+      "step": 1611
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.297037107556271e-06,
+      "loss": 0.547,
+      "step": 1612
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.285449678088369e-06,
+      "loss": 0.5553,
+      "step": 1613
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273873630956811e-06,
+      "loss": 0.5548,
+      "step": 1614
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.262308989218961e-06,
+      "loss": 0.5437,
+      "step": 1615
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.250755775909465e-06,
+      "loss": 0.5448,
+      "step": 1616
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2392140140401996e-06,
+      "loss": 0.5578,
+      "step": 1617
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.227683726600251e-06,
+      "loss": 0.5506,
+      "step": 1618
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.216164936555823e-06,
+      "loss": 0.5602,
+      "step": 1619
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204657666850245e-06,
+      "loss": 0.5643,
+      "step": 1620
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.193161940403882e-06,
+      "loss": 0.5571,
+      "step": 1621
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.181677780114113e-06,
+      "loss": 0.5642,
+      "step": 1622
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170205208855281e-06,
+      "loss": 0.5509,
+      "step": 1623
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158744249478641e-06,
+      "loss": 0.5559,
+      "step": 1624
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.147294924812332e-06,
+      "loss": 0.5453,
+      "step": 1625
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.135857257661295e-06,
+      "loss": 0.5483,
+      "step": 1626
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.124431270807277e-06,
+      "loss": 0.5414,
+      "step": 1627
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113016987008743e-06,
+      "loss": 0.5618,
+      "step": 1628
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.101614429000857e-06,
+      "loss": 0.5601,
+      "step": 1629
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.090223619495419e-06,
+      "loss": 0.5626,
+      "step": 1630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.078844581180833e-06,
+      "loss": 0.5528,
+      "step": 1631
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.067477336722063e-06,
+      "loss": 0.5436,
+      "step": 1632
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.056121908760571e-06,
+      "loss": 0.5642,
+      "step": 1633
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.044778319914288e-06,
+      "loss": 0.5591,
+      "step": 1634
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.033446592777558e-06,
+      "loss": 0.5488,
+      "step": 1635
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.022126749921116e-06,
+      "loss": 0.5524,
+      "step": 1636
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.010818813892e-06,
+      "loss": 0.5548,
+      "step": 1637
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999522807213549e-06,
+      "loss": 0.5521,
+      "step": 1638
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.988238752385341e-06,
+      "loss": 0.5527,
+      "step": 1639
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.97696667188314e-06,
+      "loss": 0.558,
+      "step": 1640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.965706588158865e-06,
+      "loss": 0.5687,
+      "step": 1641
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.954458523640535e-06,
+      "loss": 0.5598,
+      "step": 1642
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.943222500732241e-06,
+      "loss": 0.5628,
+      "step": 1643
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931998541814069e-06,
+      "loss": 0.5589,
+      "step": 1644
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.920786669242089e-06,
+      "loss": 0.5492,
+      "step": 1645
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9095869053483025e-06,
+      "loss": 0.549,
+      "step": 1646
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89839927244058e-06,
+      "loss": 0.5488,
+      "step": 1647
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.887223792802635e-06,
+      "loss": 0.5461,
+      "step": 1648
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.876060488693971e-06,
+      "loss": 0.5515,
+      "step": 1649
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.86490938234985e-06,
+      "loss": 0.5506,
+      "step": 1650
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85377049598123e-06,
+      "loss": 0.5513,
+      "step": 1651
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.84264385177472e-06,
+      "loss": 0.5439,
+      "step": 1652
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8315294718925656e-06,
+      "loss": 0.5645,
+      "step": 1653
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820427378472569e-06,
+      "loss": 0.5503,
+      "step": 1654
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8093375936280665e-06,
+      "loss": 0.5505,
+      "step": 1655
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.79826013944787e-06,
+      "loss": 0.5477,
+      "step": 1656
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7871950379962463e-06,
+      "loss": 0.5587,
+      "step": 1657
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7761423113128427e-06,
+      "loss": 0.5455,
+      "step": 1658
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.5563,
+      "step": 1659
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.754074070286027e-06,
+      "loss": 0.5577,
+      "step": 1660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7430585998985004e-06,
+      "loss": 0.5625,
+      "step": 1661
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.732055592190893e-06,
+      "loss": 0.5554,
+      "step": 1662
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.721065069079165e-06,
+      "loss": 0.5479,
+      "step": 1663
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.710087052454433e-06,
+      "loss": 0.546,
+      "step": 1664
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6991215641828903e-06,
+      "loss": 0.563,
+      "step": 1665
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.688168626105778e-06,
+      "loss": 0.5518,
+      "step": 1666
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6772282600393393e-06,
+      "loss": 0.5552,
+      "step": 1667
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6663004877747765e-06,
+      "loss": 0.5436,
+      "step": 1668
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.655385331078217e-06,
+      "loss": 0.5522,
+      "step": 1669
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6444828116906372e-06,
+      "loss": 0.5441,
+      "step": 1670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6335929513278667e-06,
+      "loss": 0.5458,
+      "step": 1671
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.622715771680508e-06,
+      "loss": 0.5523,
+      "step": 1672
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6118512944139084e-06,
+      "loss": 0.5519,
+      "step": 1673
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.600999541168113e-06,
+      "loss": 0.5585,
+      "step": 1674
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5901605335578214e-06,
+      "loss": 0.5515,
+      "step": 1675
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.579334293172357e-06,
+      "loss": 0.5671,
+      "step": 1676
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.568520841575601e-06,
+      "loss": 0.5458,
+      "step": 1677
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5577202003059664e-06,
+      "loss": 0.5609,
+      "step": 1678
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5469323908763507e-06,
+      "loss": 0.55,
+      "step": 1679
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5361574347740912e-06,
+      "loss": 0.561,
+      "step": 1680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.525395353460924e-06,
+      "loss": 0.5454,
+      "step": 1681
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5146461683729392e-06,
+      "loss": 0.5553,
+      "step": 1682
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5039099009205503e-06,
+      "loss": 0.5496,
+      "step": 1683
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4931865724884274e-06,
+      "loss": 0.542,
+      "step": 1684
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4824762044354763e-06,
+      "loss": 0.5694,
+      "step": 1685
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4717788180947855e-06,
+      "loss": 0.5603,
+      "step": 1686
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4610944347735864e-06,
+      "loss": 0.5551,
+      "step": 1687
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.450423075753213e-06,
+      "loss": 0.5546,
+      "step": 1688
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.439764762289051e-06,
+      "loss": 0.5622,
+      "step": 1689
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4291195156105118e-06,
+      "loss": 0.5516,
+      "step": 1690
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.418487356920974e-06,
+      "loss": 0.5497,
+      "step": 1691
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.407868307397747e-06,
+      "loss": 0.5536,
+      "step": 1692
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3972623881920296e-06,
+      "loss": 0.5586,
+      "step": 1693
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.386669620428865e-06,
+      "loss": 0.546,
+      "step": 1694
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.376090025207115e-06,
+      "loss": 0.5567,
+      "step": 1695
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3655236235993783e-06,
+      "loss": 0.5512,
+      "step": 1696
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.354970436652001e-06,
+      "loss": 0.5579,
+      "step": 1697
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3444304853849918e-06,
+      "loss": 0.5618,
+      "step": 1698
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3339037907920024e-06,
+      "loss": 0.5624,
+      "step": 1699
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.323390373840276e-06,
+      "loss": 0.5581,
+      "step": 1700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312890255470609e-06,
+      "loss": 0.5509,
+      "step": 1701
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3024034565973185e-06,
+      "loss": 0.5494,
+      "step": 1702
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.291929998108182e-06,
+      "loss": 0.5605,
+      "step": 1703
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2814699008644068e-06,
+      "loss": 0.5558,
+      "step": 1704
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.271023185700587e-06,
+      "loss": 0.5471,
+      "step": 1705
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.260589873424672e-06,
+      "loss": 0.5324,
+      "step": 1706
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.250169984817897e-06,
+      "loss": 0.5466,
+      "step": 1707
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2397635406347705e-06,
+      "loss": 0.568,
+      "step": 1708
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2293705616030267e-06,
+      "loss": 0.5578,
+      "step": 1709
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2189910684235705e-06,
+      "loss": 0.5645,
+      "step": 1710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2086250817704488e-06,
+      "loss": 0.5538,
+      "step": 1711
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1982726222908046e-06,
+      "loss": 0.5436,
+      "step": 1712
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.187933710604847e-06,
+      "loss": 0.5475,
+      "step": 1713
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1776083673057834e-06,
+      "loss": 0.5436,
+      "step": 1714
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.167296612959803e-06,
+      "loss": 0.5573,
+      "step": 1715
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1569984681060375e-06,
+      "loss": 0.5544,
+      "step": 1716
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1467139532564985e-06,
+      "loss": 0.5439,
+      "step": 1717
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.136443088896054e-06,
+      "loss": 0.5519,
+      "step": 1718
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1261858954823798e-06,
+      "loss": 0.5548,
+      "step": 1719
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.115942393445931e-06,
+      "loss": 0.5588,
+      "step": 1720
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1057126031898843e-06,
+      "loss": 0.5517,
+      "step": 1721
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0954965450900963e-06,
+      "loss": 0.5445,
+      "step": 1722
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0852942394950915e-06,
+      "loss": 0.5602,
+      "step": 1723
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.075105706725989e-06,
+      "loss": 0.5502,
+      "step": 1724
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.064930967076477e-06,
+      "loss": 0.5601,
+      "step": 1725
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.054770040812769e-06,
+      "loss": 0.5402,
+      "step": 1726
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0446229481735713e-06,
+      "loss": 0.5616,
+      "step": 1727
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0344897093700333e-06,
+      "loss": 0.5519,
+      "step": 1728
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0243703445856985e-06,
+      "loss": 0.5502,
+      "step": 1729
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0142648739764956e-06,
+      "loss": 0.5579,
+      "step": 1730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0041733176706668e-06,
+      "loss": 0.5521,
+      "step": 1731
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9940956957687396e-06,
+      "loss": 0.5555,
+      "step": 1732
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9840320283434865e-06,
+      "loss": 0.5602,
+      "step": 1733
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9739823354398958e-06,
+      "loss": 0.55,
+      "step": 1734
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.963946637075107e-06,
+      "loss": 0.5418,
+      "step": 1735
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.953924953238392e-06,
+      "loss": 0.5527,
+      "step": 1736
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.943917303891107e-06,
+      "loss": 0.5548,
+      "step": 1737
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9339237089666526e-06,
+      "loss": 0.5545,
+      "step": 1738
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9239441883704455e-06,
+      "loss": 0.557,
+      "step": 1739
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.913978761979849e-06,
+      "loss": 0.5549,
+      "step": 1740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9040274496441732e-06,
+      "loss": 0.5414,
+      "step": 1741
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8940902711846052e-06,
+      "loss": 0.5424,
+      "step": 1742
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8841672463941827e-06,
+      "loss": 0.569,
+      "step": 1743
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.874258395037751e-06,
+      "loss": 0.5548,
+      "step": 1744
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.864363736851922e-06,
+      "loss": 0.5561,
+      "step": 1745
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.854483291545046e-06,
+      "loss": 0.5514,
+      "step": 1746
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.844617078797155e-06,
+      "loss": 0.5439,
+      "step": 1747
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.834765118259937e-06,
+      "loss": 0.5459,
+      "step": 1748
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8249274295566863e-06,
+      "loss": 0.5576,
+      "step": 1749
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8151040322822853e-06,
+      "loss": 0.5452,
+      "step": 1750
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.80529494600313e-06,
+      "loss": 0.5613,
+      "step": 1751
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.795500190257122e-06,
+      "loss": 0.5526,
+      "step": 1752
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.785719784553624e-06,
+      "loss": 0.5479,
+      "step": 1753
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.775953748373408e-06,
+      "loss": 0.5512,
+      "step": 1754
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.766202101168628e-06,
+      "loss": 0.5456,
+      "step": 1755
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.756464862362772e-06,
+      "loss": 0.5445,
+      "step": 1756
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.746742051350646e-06,
+      "loss": 0.5511,
+      "step": 1757
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7370336874982952e-06,
+      "loss": 0.5491,
+      "step": 1758
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7273397901430023e-06,
+      "loss": 0.5588,
+      "step": 1759
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7176603785932364e-06,
+      "loss": 0.5595,
+      "step": 1760
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7079954721286108e-06,
+      "loss": 0.5479,
+      "step": 1761
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6983450899998432e-06,
+      "loss": 0.5406,
+      "step": 1762
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.688709251428725e-06,
+      "loss": 0.5595,
+      "step": 1763
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6790879756080845e-06,
+      "loss": 0.5479,
+      "step": 1764
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 0.5539,
+      "step": 1765
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.659889188844453e-06,
+      "loss": 0.5447,
+      "step": 1766
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6503117161419246e-06,
+      "loss": 0.5601,
+      "step": 1767
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6407488826707197e-06,
+      "loss": 0.5573,
+      "step": 1768
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6312007074782497e-06,
+      "loss": 0.5572,
+      "step": 1769
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6216672095827267e-06,
+      "loss": 0.5505,
+      "step": 1770
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.612148407973134e-06,
+      "loss": 0.5541,
+      "step": 1771
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.602644321609181e-06,
+      "loss": 0.5488,
+      "step": 1772
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5931549694212545e-06,
+      "loss": 0.5423,
+      "step": 1773
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.583680370310414e-06,
+      "loss": 0.5528,
+      "step": 1774
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.57422054314832e-06,
+      "loss": 0.5479,
+      "step": 1775
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.564775506777214e-06,
+      "loss": 0.5592,
+      "step": 1776
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.555345280009872e-06,
+      "loss": 0.557,
+      "step": 1777
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.545929881629585e-06,
+      "loss": 0.5452,
+      "step": 1778
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.536529330390095e-06,
+      "loss": 0.5534,
+      "step": 1779
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.527143645015577e-06,
+      "loss": 0.5544,
+      "step": 1780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5177728442005956e-06,
+      "loss": 0.5422,
+      "step": 1781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5084169466100626e-06,
+      "loss": 0.5524,
+      "step": 1782
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.499075970879222e-06,
+      "loss": 0.546,
+      "step": 1783
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4897499356135712e-06,
+      "loss": 0.5604,
+      "step": 1784
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.480438859388873e-06,
+      "loss": 0.5586,
+      "step": 1785
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4711427607510773e-06,
+      "loss": 0.5519,
+      "step": 1786
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.461861658216311e-06,
+      "loss": 0.5521,
+      "step": 1787
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.452595570270828e-06,
+      "loss": 0.5599,
+      "step": 1788
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4433445153709722e-06,
+      "loss": 0.5568,
+      "step": 1789
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4341085119431553e-06,
+      "loss": 0.5522,
+      "step": 1790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424887578383799e-06,
+      "loss": 0.5468,
+      "step": 1791
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4156817330593128e-06,
+      "loss": 0.5504,
+      "step": 1792
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.406490994306052e-06,
+      "loss": 0.5495,
+      "step": 1793
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.397315380430284e-06,
+      "loss": 0.546,
+      "step": 1794
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3881549097081467e-06,
+      "loss": 0.5621,
+      "step": 1795
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3790096003856178e-06,
+      "loss": 0.5425,
+      "step": 1796
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.36987947067848e-06,
+      "loss": 0.5547,
+      "step": 1797
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3607645387722753e-06,
+      "loss": 0.5484,
+      "step": 1798
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.351664822822277e-06,
+      "loss": 0.5511,
+      "step": 1799
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342580340953451e-06,
+      "loss": 0.5512,
+      "step": 1800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3335111112604194e-06,
+      "loss": 0.5537,
+      "step": 1801
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3244571518074223e-06,
+      "loss": 0.5604,
+      "step": 1802
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3154184806282863e-06,
+      "loss": 0.557,
+      "step": 1803
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.306395115726392e-06,
+      "loss": 0.5594,
+      "step": 1804
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2973870750746253e-06,
+      "loss": 0.5505,
+      "step": 1805
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2883943766153503e-06,
+      "loss": 0.5444,
+      "step": 1806
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.279417038260373e-06,
+      "loss": 0.5606,
+      "step": 1807
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.270455077890904e-06,
+      "loss": 0.5528,
+      "step": 1808
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.261508513357532e-06,
+      "loss": 0.549,
+      "step": 1809
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2525773624801627e-06,
+      "loss": 0.5574,
+      "step": 1810
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2436616430480197e-06,
+      "loss": 0.5547,
+      "step": 1811
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.234761372819577e-06,
+      "loss": 0.5659,
+      "step": 1812
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2258765695225416e-06,
+      "loss": 0.5515,
+      "step": 1813
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.217007250853812e-06,
+      "loss": 0.5561,
+      "step": 1814
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.208153434479442e-06,
+      "loss": 0.543,
+      "step": 1815
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.199315138034619e-06,
+      "loss": 0.5423,
+      "step": 1816
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1904923791235965e-06,
+      "loss": 0.5541,
+      "step": 1817
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.5545,
+      "step": 1818
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1728935441652687e-06,
+      "loss": 0.5605,
+      "step": 1819
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1641175031716112e-06,
+      "loss": 0.5453,
+      "step": 1820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.155357069818995e-06,
+      "loss": 0.5479,
+      "step": 1821
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1466122615565945e-06,
+      "loss": 0.5546,
+      "step": 1822
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.137883095802469e-06,
+      "loss": 0.5468,
+      "step": 1823
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.129169589943514e-06,
+      "loss": 0.5593,
+      "step": 1824
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.120471761335434e-06,
+      "loss": 0.5541,
+      "step": 1825
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.111789627302707e-06,
+      "loss": 0.5554,
+      "step": 1826
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1031232051385606e-06,
+      "loss": 0.555,
+      "step": 1827
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0944725121049104e-06,
+      "loss": 0.5482,
+      "step": 1828
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.085837565432349e-06,
+      "loss": 0.5561,
+      "step": 1829
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.077218382320113e-06,
+      "loss": 0.5433,
+      "step": 1830
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.068614979936032e-06,
+      "loss": 0.544,
+      "step": 1831
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.060027375416507e-06,
+      "loss": 0.5444,
+      "step": 1832
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0514555858664663e-06,
+      "loss": 0.5431,
+      "step": 1833
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0428996283593506e-06,
+      "loss": 0.5632,
+      "step": 1834
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.034359519937057e-06,
+      "loss": 0.5568,
+      "step": 1835
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.025835277609908e-06,
+      "loss": 0.5606,
+      "step": 1836
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.017326918356639e-06,
+      "loss": 0.5605,
+      "step": 1837
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.008834459124338e-06,
+      "loss": 0.5474,
+      "step": 1838
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.000357916828428e-06,
+      "loss": 0.5536,
+      "step": 1839
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.991897308352624e-06,
+      "loss": 0.5588,
+      "step": 1840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9834526505489105e-06,
+      "loss": 0.5591,
+      "step": 1841
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.975023960237499e-06,
+      "loss": 0.5454,
+      "step": 1842
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.966611254206785e-06,
+      "loss": 0.5564,
+      "step": 1843
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9582145492133463e-06,
+      "loss": 0.5521,
+      "step": 1844
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.949833861981877e-06,
+      "loss": 0.5581,
+      "step": 1845
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9414692092051666e-06,
+      "loss": 0.555,
+      "step": 1846
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93312060754407e-06,
+      "loss": 0.5533,
+      "step": 1847
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.924788073627474e-06,
+      "loss": 0.5441,
+      "step": 1848
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.916471624052256e-06,
+      "loss": 0.5519,
+      "step": 1849
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9081712753832603e-06,
+      "loss": 0.5493,
+      "step": 1850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8998870441532569e-06,
+      "loss": 0.5537,
+      "step": 1851
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.891618946862913e-06,
+      "loss": 0.5536,
+      "step": 1852
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8833669999807723e-06,
+      "loss": 0.5473,
+      "step": 1853
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.875131219943187e-06,
+      "loss": 0.5518,
+      "step": 1854
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8669116231543294e-06,
+      "loss": 0.5496,
+      "step": 1855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8587082259861267e-06,
+      "loss": 0.5493,
+      "step": 1856
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8505210447782418e-06,
+      "loss": 0.5436,
+      "step": 1857
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8423500958380369e-06,
+      "loss": 0.5497,
+      "step": 1858
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8341953954405434e-06,
+      "loss": 0.5579,
+      "step": 1859
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8260569598284361e-06,
+      "loss": 0.5606,
+      "step": 1860
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.817934805211976e-06,
+      "loss": 0.5412,
+      "step": 1861
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8098289477690135e-06,
+      "loss": 0.5558,
+      "step": 1862
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8017394036449276e-06,
+      "loss": 0.56,
+      "step": 1863
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7936661889526063e-06,
+      "loss": 0.552,
+      "step": 1864
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7856093197724133e-06,
+      "loss": 0.5615,
+      "step": 1865
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7775688121521518e-06,
+      "loss": 0.548,
+      "step": 1866
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7695446821070438e-06,
+      "loss": 0.5465,
+      "step": 1867
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.761536945619684e-06,
+      "loss": 0.5491,
+      "step": 1868
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7535456186400123e-06,
+      "loss": 0.5646,
+      "step": 1869
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7455707170852876e-06,
+      "loss": 0.5499,
+      "step": 1870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 0.5641,
+      "step": 1871
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7296702537560994e-06,
+      "loss": 0.5511,
+      "step": 1872
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.72174472365244e-06,
+      "loss": 0.5551,
+      "step": 1873
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7138356823152825e-06,
+      "loss": 0.5492,
+      "step": 1874
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7059431454979825e-06,
+      "loss": 0.5483,
+      "step": 1875
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6980671289210281e-06,
+      "loss": 0.5452,
+      "step": 1876
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6902076482719987e-06,
+      "loss": 0.5479,
+      "step": 1877
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.682364719205537e-06,
+      "loss": 0.5466,
+      "step": 1878
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.674538357343326e-06,
+      "loss": 0.5528,
+      "step": 1879
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.666728578274035e-06,
+      "loss": 0.5407,
+      "step": 1880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6589353975533174e-06,
+      "loss": 0.5531,
+      "step": 1881
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6511588307037596e-06,
+      "loss": 0.5591,
+      "step": 1882
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6433988932148547e-06,
+      "loss": 0.5611,
+      "step": 1883
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6356556005429758e-06,
+      "loss": 0.5445,
+      "step": 1884
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6279289681113407e-06,
+      "loss": 0.5508,
+      "step": 1885
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6202190113099914e-06,
+      "loss": 0.549,
+      "step": 1886
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6125257454957365e-06,
+      "loss": 0.5521,
+      "step": 1887
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.60484918599216e-06,
+      "loss": 0.5523,
+      "step": 1888
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5971893480895583e-06,
+      "loss": 0.559,
+      "step": 1889
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.589546247044923e-06,
+      "loss": 0.5532,
+      "step": 1890
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5819198980819096e-06,
+      "loss": 0.554,
+      "step": 1891
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5743103163908048e-06,
+      "loss": 0.5549,
+      "step": 1892
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5667175171285054e-06,
+      "loss": 0.5577,
+      "step": 1893
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.559141515418472e-06,
+      "loss": 0.5617,
+      "step": 1894
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5515823263507112e-06,
+      "loss": 0.5485,
+      "step": 1895
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5440399649817384e-06,
+      "loss": 0.546,
+      "step": 1896
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5365144463345627e-06,
+      "loss": 0.552,
+      "step": 1897
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5290057853986262e-06,
+      "loss": 0.5408,
+      "step": 1898
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5215139971298131e-06,
+      "loss": 0.5666,
+      "step": 1899
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5140390964503881e-06,
+      "loss": 0.5503,
+      "step": 1900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5065810982489849e-06,
+      "loss": 0.5449,
+      "step": 1901
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.499140017380566e-06,
+      "loss": 0.5569,
+      "step": 1902
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4917158686663992e-06,
+      "loss": 0.5523,
+      "step": 1903
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4843086668940331e-06,
+      "loss": 0.5517,
+      "step": 1904
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4769184268172465e-06,
+      "loss": 0.5597,
+      "step": 1905
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.469545163156051e-06,
+      "loss": 0.556,
+      "step": 1906
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4621888905966308e-06,
+      "loss": 0.5416,
+      "step": 1907
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4548496237913335e-06,
+      "loss": 0.5475,
+      "step": 1908
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4475273773586319e-06,
+      "loss": 0.5461,
+      "step": 1909
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4402221658830963e-06,
+      "loss": 0.5458,
+      "step": 1910
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4329340039153738e-06,
+      "loss": 0.5562,
+      "step": 1911
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4256629059721427e-06,
+      "loss": 0.5566,
+      "step": 1912
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4184088865360978e-06,
+      "loss": 0.5497,
+      "step": 1913
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4111719600559159e-06,
+      "loss": 0.5587,
+      "step": 1914
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4039521409462265e-06,
+      "loss": 0.5515,
+      "step": 1915
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3967494435875861e-06,
+      "loss": 0.5545,
+      "step": 1916
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3895638823264447e-06,
+      "loss": 0.5494,
+      "step": 1917
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3823954714751265e-06,
+      "loss": 0.5562,
+      "step": 1918
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3752442253117903e-06,
+      "loss": 0.5501,
+      "step": 1919
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3681101580804057e-06,
+      "loss": 0.5501,
+      "step": 1920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3609932839907281e-06,
+      "loss": 0.5489,
+      "step": 1921
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3538936172182627e-06,
+      "loss": 0.5508,
+      "step": 1922
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3468111719042497e-06,
+      "loss": 0.5572,
+      "step": 1923
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.5567,
+      "step": 1924
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3326980020449621e-06,
+      "loss": 0.5617,
+      "step": 1925
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3256673056105385e-06,
+      "loss": 0.5548,
+      "step": 1926
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3186538868562004e-06,
+      "loss": 0.5485,
+      "step": 1927
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3116577597513924e-06,
+      "loss": 0.5444,
+      "step": 1928
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3046789382311132e-06,
+      "loss": 0.5437,
+      "step": 1929
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2977174361959022e-06,
+      "loss": 0.5449,
+      "step": 1930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2907732675117878e-06,
+      "loss": 0.5605,
+      "step": 1931
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2838464460102862e-06,
+      "loss": 0.5426,
+      "step": 1932
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2769369854883528e-06,
+      "loss": 0.5456,
+      "step": 1933
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.270044899708367e-06,
+      "loss": 0.5492,
+      "step": 1934
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2631702023980997e-06,
+      "loss": 0.5555,
+      "step": 1935
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2563129072506852e-06,
+      "loss": 0.543,
+      "step": 1936
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2494730279246014e-06,
+      "loss": 0.5571,
+      "step": 1937
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2426505780436326e-06,
+      "loss": 0.5513,
+      "step": 1938
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2358455711968463e-06,
+      "loss": 0.5514,
+      "step": 1939
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2290580209385684e-06,
+      "loss": 0.5462,
+      "step": 1940
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2222879407883592e-06,
+      "loss": 0.558,
+      "step": 1941
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2155353442309714e-06,
+      "loss": 0.557,
+      "step": 1942
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2088002447163383e-06,
+      "loss": 0.5524,
+      "step": 1943
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2020826556595477e-06,
+      "loss": 0.5554,
+      "step": 1944
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1953825904408033e-06,
+      "loss": 0.5553,
+      "step": 1945
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1887000624054056e-06,
+      "loss": 0.5587,
+      "step": 1946
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.182035084863724e-06,
+      "loss": 0.5509,
+      "step": 1947
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1753876710911772e-06,
+      "loss": 0.5529,
+      "step": 1948
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.168757834328188e-06,
+      "loss": 0.5437,
+      "step": 1949
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1621455877801757e-06,
+      "loss": 0.5424,
+      "step": 1950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1555509446175284e-06,
+      "loss": 0.5411,
+      "step": 1951
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1489739179755622e-06,
+      "loss": 0.5442,
+      "step": 1952
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1424145209545079e-06,
+      "loss": 0.5566,
+      "step": 1953
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1358727666194812e-06,
+      "loss": 0.549,
+      "step": 1954
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1293486680004607e-06,
+      "loss": 0.5587,
+      "step": 1955
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1228422380922533e-06,
+      "loss": 0.5403,
+      "step": 1956
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1163534898544692e-06,
+      "loss": 0.5486,
+      "step": 1957
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1098824362115102e-06,
+      "loss": 0.5477,
+      "step": 1958
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1034290900525279e-06,
+      "loss": 0.5457,
+      "step": 1959
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.096993464231404e-06,
+      "loss": 0.543,
+      "step": 1960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0905755715667222e-06,
+      "loss": 0.5436,
+      "step": 1961
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0841754248417535e-06,
+      "loss": 0.5553,
+      "step": 1962
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0777930368044143e-06,
+      "loss": 0.5584,
+      "step": 1963
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0714284201672475e-06,
+      "loss": 0.5606,
+      "step": 1964
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.065081587607406e-06,
+      "loss": 0.537,
+      "step": 1965
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0587525517666142e-06,
+      "loss": 0.5557,
+      "step": 1966
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0524413252511567e-06,
+      "loss": 0.5596,
+      "step": 1967
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0461479206318303e-06,
+      "loss": 0.566,
+      "step": 1968
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0398723504439512e-06,
+      "loss": 0.5561,
+      "step": 1969
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0336146271872993e-06,
+      "loss": 0.5405,
+      "step": 1970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0273747633261144e-06,
+      "loss": 0.555,
+      "step": 1971
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0211527712890601e-06,
+      "loss": 0.5503,
+      "step": 1972
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0149486634692019e-06,
+      "loss": 0.5609,
+      "step": 1973
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0087624522239902e-06,
+      "loss": 0.5538,
+      "step": 1974
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0025941498752167e-06,
+      "loss": 0.5592,
+      "step": 1975
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.964437687090124e-07,
+      "loss": 0.5312,
+      "step": 1976
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.5498,
+      "step": 1977
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.841968188903173e-07,
+      "loss": 0.5644,
+      "step": 1978
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.781002746315039e-07,
+      "loss": 0.5537,
+      "step": 1979
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.720217003425648e-07,
+      "loss": 0.5496,
+      "step": 1980
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.659611081309095e-07,
+      "loss": 0.5526,
+      "step": 1981
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.59918510068124e-07,
+      "loss": 0.5532,
+      "step": 1982
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.538939181899565e-07,
+      "loss": 0.5435,
+      "step": 1983
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.478873444962888e-07,
+      "loss": 0.5535,
+      "step": 1984
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.418988009511143e-07,
+      "loss": 0.5395,
+      "step": 1985
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.359282994825125e-07,
+      "loss": 0.5639,
+      "step": 1986
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.299758519826274e-07,
+      "loss": 0.5435,
+      "step": 1987
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.240414703076461e-07,
+      "loss": 0.5541,
+      "step": 1988
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.181251662777668e-07,
+      "loss": 0.5408,
+      "step": 1989
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.12226951677182e-07,
+      "loss": 0.5417,
+      "step": 1990
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.06346838254053e-07,
+      "loss": 0.5478,
+      "step": 1991
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.004848377204878e-07,
+      "loss": 0.5457,
+      "step": 1992
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.946409617525175e-07,
+      "loss": 0.5467,
+      "step": 1993
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.888152219900692e-07,
+      "loss": 0.5501,
+      "step": 1994
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.830076300369517e-07,
+      "loss": 0.5512,
+      "step": 1995
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.772181974608207e-07,
+      "loss": 0.5512,
+      "step": 1996
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.714469357931654e-07,
+      "loss": 0.5562,
+      "step": 1997
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.656938565292805e-07,
+      "loss": 0.5602,
+      "step": 1998
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.599589711282419e-07,
+      "loss": 0.5456,
+      "step": 1999
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.542422910128967e-07,
+      "loss": 0.5543,
+      "step": 2000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.485438275698154e-07,
+      "loss": 0.5578,
+      "step": 2001
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.428635921492978e-07,
+      "loss": 0.547,
+      "step": 2002
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.37201596065329e-07,
+      "loss": 0.5547,
+      "step": 2003
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.315578505955679e-07,
+      "loss": 0.5532,
+      "step": 2004
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.259323669813202e-07,
+      "loss": 0.5582,
+      "step": 2005
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.20325156427515e-07,
+      "loss": 0.5528,
+      "step": 2006
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.14736230102694e-07,
+      "loss": 0.5503,
+      "step": 2007
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.091655991389668e-07,
+      "loss": 0.5556,
+      "step": 2008
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.036132746320125e-07,
+      "loss": 0.5604,
+      "step": 2009
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.980792676410409e-07,
+      "loss": 0.5519,
+      "step": 2010
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.925635891887839e-07,
+      "loss": 0.544,
+      "step": 2011
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.870662502614568e-07,
+      "loss": 0.5531,
+      "step": 2012
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.815872618087506e-07,
+      "loss": 0.5389,
+      "step": 2013
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.76126634743809e-07,
+      "loss": 0.5592,
+      "step": 2014
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.706843799431985e-07,
+      "loss": 0.5366,
+      "step": 2015
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.652605082468912e-07,
+      "loss": 0.5394,
+      "step": 2016
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.598550304582453e-07,
+      "loss": 0.5529,
+      "step": 2017
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.54467957343985e-07,
+      "loss": 0.5533,
+      "step": 2018
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.490992996341662e-07,
+      "loss": 0.5476,
+      "step": 2019
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.437490680221737e-07,
+      "loss": 0.5418,
+      "step": 2020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.384172731646877e-07,
+      "loss": 0.5456,
+      "step": 2021
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.331039256816664e-07,
+      "loss": 0.5352,
+      "step": 2022
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.278090361563228e-07,
+      "loss": 0.5449,
+      "step": 2023
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.225326151351042e-07,
+      "loss": 0.5545,
+      "step": 2024
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.17274673127677e-07,
+      "loss": 0.5548,
+      "step": 2025
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.120352206068981e-07,
+      "loss": 0.5583,
+      "step": 2026
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.068142680087909e-07,
+      "loss": 0.5461,
+      "step": 2027
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.016118257325399e-07,
+      "loss": 0.5531,
+      "step": 2028
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.964279041404553e-07,
+      "loss": 0.5374,
+      "step": 2029
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.912625135579587e-07,
+      "loss": 0.5518,
+      "step": 2030
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.861156642735578e-07,
+      "loss": 0.5534,
+      "step": 2031
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.809873665388367e-07,
+      "loss": 0.5522,
+      "step": 2032
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.758776305684245e-07,
+      "loss": 0.5438,
+      "step": 2033
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.707864665399733e-07,
+      "loss": 0.5492,
+      "step": 2034
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.657138845941524e-07,
+      "loss": 0.5426,
+      "step": 2035
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.606598948346132e-07,
+      "loss": 0.5434,
+      "step": 2036
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.556245073279777e-07,
+      "loss": 0.5517,
+      "step": 2037
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.506077321038118e-07,
+      "loss": 0.557,
+      "step": 2038
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.456095791546147e-07,
+      "loss": 0.5581,
+      "step": 2039
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.406300584357905e-07,
+      "loss": 0.5431,
+      "step": 2040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.3566917986563e-07,
+      "loss": 0.545,
+      "step": 2041
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.307269533252946e-07,
+      "loss": 0.5516,
+      "step": 2042
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.258033886587911e-07,
+      "loss": 0.5653,
+      "step": 2043
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.208984956729613e-07,
+      "loss": 0.549,
+      "step": 2044
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.160122841374482e-07,
+      "loss": 0.5567,
+      "step": 2045
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.111447637846923e-07,
+      "loss": 0.5452,
+      "step": 2046
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.062959443099014e-07,
+      "loss": 0.551,
+      "step": 2047
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.014658353710334e-07,
+      "loss": 0.5566,
+      "step": 2048
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.966544465887803e-07,
+      "loss": 0.5414,
+      "step": 2049
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.918617875465449e-07,
+      "loss": 0.5355,
+      "step": 2050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.870878677904302e-07,
+      "loss": 0.5485,
+      "step": 2051
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.823326968292009e-07,
+      "loss": 0.5396,
+      "step": 2052
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.775962841342919e-07,
+      "loss": 0.5449,
+      "step": 2053
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.728786391397668e-07,
+      "loss": 0.5431,
+      "step": 2054
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.681797712423099e-07,
+      "loss": 0.5474,
+      "step": 2055
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.634996898012024e-07,
+      "loss": 0.551,
+      "step": 2056
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.588384041383089e-07,
+      "loss": 0.5512,
+      "step": 2057
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.541959235380578e-07,
+      "loss": 0.5457,
+      "step": 2058
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.495722572474183e-07,
+      "loss": 0.5597,
+      "step": 2059
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.449674144758854e-07,
+      "loss": 0.5518,
+      "step": 2060
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.403814043954592e-07,
+      "loss": 0.5474,
+      "step": 2061
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.358142361406371e-07,
+      "loss": 0.5507,
+      "step": 2062
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.312659188083746e-07,
+      "loss": 0.5485,
+      "step": 2063
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.267364614580861e-07,
+      "loss": 0.5487,
+      "step": 2064
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.222258731116237e-07,
+      "loss": 0.5496,
+      "step": 2065
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.177341627532506e-07,
+      "loss": 0.558,
+      "step": 2066
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.132613393296293e-07,
+      "loss": 0.5368,
+      "step": 2067
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.088074117498043e-07,
+      "loss": 0.5544,
+      "step": 2068
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.043723888851837e-07,
+      "loss": 0.5652,
+      "step": 2069
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.999562795695201e-07,
+      "loss": 0.5381,
+      "step": 2070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.955590925988896e-07,
+      "loss": 0.5527,
+      "step": 2071
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.911808367316861e-07,
+      "loss": 0.54,
+      "step": 2072
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.868215206885918e-07,
+      "loss": 0.5411,
+      "step": 2073
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.824811531525642e-07,
+      "loss": 0.551,
+      "step": 2074
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.781597427688189e-07,
+      "loss": 0.5619,
+      "step": 2075
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.738572981448153e-07,
+      "loss": 0.5491,
+      "step": 2076
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.695738278502338e-07,
+      "loss": 0.5372,
+      "step": 2077
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.653093404169573e-07,
+      "loss": 0.5432,
+      "step": 2078
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.61063844339068e-07,
+      "loss": 0.5453,
+      "step": 2079
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.568373480728139e-07,
+      "loss": 0.5468,
+      "step": 2080
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.526298600365997e-07,
+      "loss": 0.5576,
+      "step": 2081
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4844138861096954e-07,
+      "loss": 0.548,
+      "step": 2082
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 0.5429,
+      "step": 2083
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.401215289242411e-07,
+      "loss": 0.5508,
+      "step": 2084
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.359901572347758e-07,
+      "loss": 0.5562,
+      "step": 2085
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.318778352991326e-07,
+      "loss": 0.552,
+      "step": 2086
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.277845713083018e-07,
+      "loss": 0.5543,
+      "step": 2087
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2371037341531805e-07,
+      "loss": 0.55,
+      "step": 2088
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.196552497352302e-07,
+      "loss": 0.5549,
+      "step": 2089
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1561920834510605e-07,
+      "loss": 0.5387,
+      "step": 2090
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.116022572839984e-07,
+      "loss": 0.5573,
+      "step": 2091
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0760440455293703e-07,
+      "loss": 0.5524,
+      "step": 2092
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.036256581149123e-07,
+      "loss": 0.5452,
+      "step": 2093
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.996660258948548e-07,
+      "loss": 0.556,
+      "step": 2094
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9572551577963135e-07,
+      "loss": 0.5537,
+      "step": 2095
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9180413561800913e-07,
+      "loss": 0.5556,
+      "step": 2096
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.879018932206624e-07,
+      "loss": 0.5396,
+      "step": 2097
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.840187963601405e-07,
+      "loss": 0.5644,
+      "step": 2098
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.801548527708621e-07,
+      "loss": 0.5465,
+      "step": 2099
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7631007014909294e-07,
+      "loss": 0.5507,
+      "step": 2100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7248445615293506e-07,
+      "loss": 0.5413,
+      "step": 2101
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.686780184023098e-07,
+      "loss": 0.5446,
+      "step": 2102
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6489076447894456e-07,
+      "loss": 0.5287,
+      "step": 2103
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6112270192635414e-07,
+      "loss": 0.5416,
+      "step": 2104
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.573738382498271e-07,
+      "loss": 0.5485,
+      "step": 2105
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5364418091641374e-07,
+      "loss": 0.5533,
+      "step": 2106
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.499337373549072e-07,
+      "loss": 0.5414,
+      "step": 2107
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4624251495582906e-07,
+      "loss": 0.5512,
+      "step": 2108
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.425705210714192e-07,
+      "loss": 0.5442,
+      "step": 2109
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3891776301561596e-07,
+      "loss": 0.5467,
+      "step": 2110
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.352842480640439e-07,
+      "loss": 0.5479,
+      "step": 2111
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.316699834539983e-07,
+      "loss": 0.5443,
+      "step": 2112
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.280749763844293e-07,
+      "loss": 0.5456,
+      "step": 2113
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2449923401593587e-07,
+      "loss": 0.5468,
+      "step": 2114
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2094276347073626e-07,
+      "loss": 0.5409,
+      "step": 2115
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1740557183267183e-07,
+      "loss": 0.552,
+      "step": 2116
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.138876661471779e-07,
+      "loss": 0.5511,
+      "step": 2117
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.103890534212783e-07,
+      "loss": 0.551,
+      "step": 2118
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.069097406235666e-07,
+      "loss": 0.5446,
+      "step": 2119
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.034497346841958e-07,
+      "loss": 0.5493,
+      "step": 2120
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.000090424948665e-07,
+      "loss": 0.5508,
+      "step": 2121
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9658767090880225e-07,
+      "loss": 0.5494,
+      "step": 2122
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.931856267407507e-07,
+      "loss": 0.5515,
+      "step": 2123
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.8980291676695917e-07,
+      "loss": 0.5465,
+      "step": 2124
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.864395477251658e-07,
+      "loss": 0.5569,
+      "step": 2125
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.83095526314584e-07,
+      "loss": 0.5524,
+      "step": 2126
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7977085919589253e-07,
+      "loss": 0.5553,
+      "step": 2127
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.764655529912175e-07,
+      "loss": 0.5381,
+      "step": 2128
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7317961428412475e-07,
+      "loss": 0.5501,
+      "step": 2129
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6991304961959895e-07,
+      "loss": 0.5434,
+      "step": 2130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6666586550403884e-07,
+      "loss": 0.544,
+      "step": 2131
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6343806840524286e-07,
+      "loss": 0.5513,
+      "step": 2132
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.602296647523861e-07,
+      "loss": 0.5522,
+      "step": 2133
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.570406609360221e-07,
+      "loss": 0.5446,
+      "step": 2134
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.538710633080621e-07,
+      "loss": 0.5437,
+      "step": 2135
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.5569,
+      "step": 2136
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.475901118317181e-07,
+      "loss": 0.545,
+      "step": 2137
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4447877049383563e-07,
+      "loss": 0.5358,
+      "step": 2138
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.413868603653413e-07,
+      "loss": 0.5468,
+      "step": 2139
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3831438760475e-07,
+      "loss": 0.5396,
+      "step": 2140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3526135833186527e-07,
+      "loss": 0.5394,
+      "step": 2141
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3222777862776046e-07,
+      "loss": 0.5454,
+      "step": 2142
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2921365453477229e-07,
+      "loss": 0.5494,
+      "step": 2143
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2621899205648169e-07,
+      "loss": 0.5392,
+      "step": 2144
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2324379715770728e-07,
+      "loss": 0.5529,
+      "step": 2145
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2028807576449208e-07,
+      "loss": 0.5565,
+      "step": 2146
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.173518337640923e-07,
+      "loss": 0.5478,
+      "step": 2147
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1443507700495968e-07,
+      "loss": 0.5533,
+      "step": 2148
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1153781129674367e-07,
+      "loss": 0.5572,
+      "step": 2149
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.086600424102625e-07,
+      "loss": 0.5472,
+      "step": 2150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0580177607750663e-07,
+      "loss": 0.5427,
+      "step": 2151
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0296301799161644e-07,
+      "loss": 0.5446,
+      "step": 2152
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.00143773806879e-07,
+      "loss": 0.5535,
+      "step": 2153
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.973440491387124e-07,
+      "loss": 0.5466,
+      "step": 2154
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9456384956365149e-07,
+      "loss": 0.5539,
+      "step": 2155
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9180318061934655e-07,
+      "loss": 0.5557,
+      "step": 2156
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.890620478045435e-07,
+      "loss": 0.5529,
+      "step": 2157
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8634045657907717e-07,
+      "loss": 0.5395,
+      "step": 2158
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8363841236385571e-07,
+      "loss": 0.5487,
+      "step": 2159
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8095592054085843e-07,
+      "loss": 0.5479,
+      "step": 2160
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7829298645311688e-07,
+      "loss": 0.5508,
+      "step": 2161
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7564961540470492e-07,
+      "loss": 0.5554,
+      "step": 2162
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7302581266073537e-07,
+      "loss": 0.5439,
+      "step": 2163
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7042158344733995e-07,
+      "loss": 0.5462,
+      "step": 2164
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6783693295166935e-07,
+      "loss": 0.5347,
+      "step": 2165
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6527186632186997e-07,
+      "loss": 0.5437,
+      "step": 2166
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.627263886670849e-07,
+      "loss": 0.5556,
+      "step": 2167
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6020050505744066e-07,
+      "loss": 0.5551,
+      "step": 2168
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5769422052403172e-07,
+      "loss": 0.5527,
+      "step": 2169
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5520754005891924e-07,
+      "loss": 0.5569,
+      "step": 2170
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5274046861511348e-07,
+      "loss": 0.5398,
+      "step": 2171
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5029301110656923e-07,
+      "loss": 0.5465,
+      "step": 2172
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4786517240817255e-07,
+      "loss": 0.5416,
+      "step": 2173
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4545695735573296e-07,
+      "loss": 0.5572,
+      "step": 2174
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4306837074597235e-07,
+      "loss": 0.5455,
+      "step": 2175
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4069941733651948e-07,
+      "loss": 0.552,
+      "step": 2176
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3835010184589325e-07,
+      "loss": 0.5413,
+      "step": 2177
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3602042895349942e-07,
+      "loss": 0.5416,
+      "step": 2178
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.337104032996206e-07,
+      "loss": 0.5414,
+      "step": 2179
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.314200294854029e-07,
+      "loss": 0.5508,
+      "step": 2180
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2914931207285154e-07,
+      "loss": 0.5487,
+      "step": 2181
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.268982555848197e-07,
+      "loss": 0.5506,
+      "step": 2182
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2466686450499866e-07,
+      "loss": 0.5529,
+      "step": 2183
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2245514327790863e-07,
+      "loss": 0.5465,
+      "step": 2184
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2026309630889465e-07,
+      "loss": 0.5434,
+      "step": 2185
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1809072796411303e-07,
+      "loss": 0.5409,
+      "step": 2186
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1593804257052143e-07,
+      "loss": 0.5575,
+      "step": 2187
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1380504441587448e-07,
+      "loss": 0.5495,
+      "step": 2188
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 0.552,
+      "step": 2189
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0959812677835968e-07,
+      "loss": 0.5604,
+      "step": 2190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0752421567490123e-07,
+      "loss": 0.5608,
+      "step": 2191
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0547000856918733e-07,
+      "loss": 0.5481,
+      "step": 2192
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0343550955282278e-07,
+      "loss": 0.5499,
+      "step": 2193
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0142072267815606e-07,
+      "loss": 0.5437,
+      "step": 2194
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.94256519582748e-08,
+      "loss": 0.5502,
+      "step": 2195
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.74503013669914e-08,
+      "loss": 0.5514,
+      "step": 2196
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.549467483884412e-08,
+      "loss": 0.5487,
+      "step": 2197
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.355877626908261e-08,
+      "loss": 0.5472,
+      "step": 2198
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.164260951366021e-08,
+      "loss": 0.553,
+      "step": 2199
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.97461783892306e-08,
+      "loss": 0.5477,
+      "step": 2200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.786948667313667e-08,
+      "loss": 0.5526,
+      "step": 2201
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.601253810340493e-08,
+      "loss": 0.549,
+      "step": 2202
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.417533637873454e-08,
+      "loss": 0.5475,
+      "step": 2203
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.235788515849607e-08,
+      "loss": 0.5482,
+      "step": 2204
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.056018806271937e-08,
+      "loss": 0.5494,
+      "step": 2205
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.878224867208573e-08,
+      "loss": 0.5489,
+      "step": 2206
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.70240705279257e-08,
+      "loss": 0.5457,
+      "step": 2207
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.528565713220803e-08,
+      "loss": 0.5507,
+      "step": 2208
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.35670119475329e-08,
+      "loss": 0.5564,
+      "step": 2209
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.186813839712537e-08,
+      "loss": 0.5486,
+      "step": 2210
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.018903986483083e-08,
+      "loss": 0.5451,
+      "step": 2211
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.852971969510514e-08,
+      "loss": 0.5474,
+      "step": 2212
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.689018119301227e-08,
+      "loss": 0.5662,
+      "step": 2213
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.527042762420888e-08,
+      "loss": 0.5537,
+      "step": 2214
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.367046221494866e-08,
+      "loss": 0.5488,
+      "step": 2215
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.209028815207019e-08,
+      "loss": 0.5498,
+      "step": 2216
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.052990858298801e-08,
+      "loss": 0.5415,
+      "step": 2217
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.89893266156949e-08,
+      "loss": 0.5483,
+      "step": 2218
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.746854531874624e-08,
+      "loss": 0.5563,
+      "step": 2219
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5967567721258996e-08,
+      "loss": 0.5534,
+      "step": 2220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4486396812906125e-08,
+      "loss": 0.5455,
+      "step": 2221
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3025035543907696e-08,
+      "loss": 0.5562,
+      "step": 2222
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.158348682502756e-08,
+      "loss": 0.5547,
+      "step": 2223
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.01617535275678e-08,
+      "loss": 0.5435,
+      "step": 2224
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8759838483358745e-08,
+      "loss": 0.5357,
+      "step": 2225
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.737774448475896e-08,
+      "loss": 0.5528,
+      "step": 2226
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6015474284646366e-08,
+      "loss": 0.561,
+      "step": 2227
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.46730305964127e-08,
+      "loss": 0.5527,
+      "step": 2228
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.335041609396018e-08,
+      "loss": 0.5435,
+      "step": 2229
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.204763341169371e-08,
+      "loss": 0.5487,
+      "step": 2230
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.076468514451759e-08,
+      "loss": 0.5477,
+      "step": 2231
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.950157384783104e-08,
+      "loss": 0.5484,
+      "step": 2232
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8258302037518234e-08,
+      "loss": 0.5526,
+      "step": 2233
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.703487218994939e-08,
+      "loss": 0.537,
+      "step": 2234
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5831286741973006e-08,
+      "loss": 0.5573,
+      "step": 2235
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.46475480909092e-08,
+      "loss": 0.554,
+      "step": 2236
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3483658594548606e-08,
+      "loss": 0.547,
+      "step": 2237
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2339620571146814e-08,
+      "loss": 0.5584,
+      "step": 2238
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.12154362994177e-08,
+      "loss": 0.5471,
+      "step": 2239
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.011110801852901e-08,
+      "loss": 0.5458,
+      "step": 2240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.902663792810012e-08,
+      "loss": 0.5567,
+      "step": 2241
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.796202818819871e-08,
+      "loss": 0.558,
+      "step": 2242
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6917280919329656e-08,
+      "loss": 0.5531,
+      "step": 2243
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5892398202439494e-08,
+      "loss": 0.5476,
+      "step": 2244
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4887382078905287e-08,
+      "loss": 0.5339,
+      "step": 2245
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3902234550536862e-08,
+      "loss": 0.5569,
+      "step": 2246
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.293695757956571e-08,
+      "loss": 0.55,
+      "step": 2247
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.199155308864609e-08,
+      "loss": 0.5483,
+      "step": 2248
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1066022960852806e-08,
+      "loss": 0.5477,
+      "step": 2249
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0160369039670115e-08,
+      "loss": 0.5579,
+      "step": 2250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9274593128996155e-08,
+      "loss": 0.5486,
+      "step": 2251
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.840869699313186e-08,
+      "loss": 0.5588,
+      "step": 2252
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7562682356786488e-08,
+      "loss": 0.5408,
+      "step": 2253
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.673655090506321e-08,
+      "loss": 0.5634,
+      "step": 2254
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.593030428346576e-08,
+      "loss": 0.5416,
+      "step": 2255
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.514394409788955e-08,
+      "loss": 0.5519,
+      "step": 2256
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4377471914619468e-08,
+      "loss": 0.555,
+      "step": 2257
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.363088926032763e-08,
+      "loss": 0.5402,
+      "step": 2258
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.290419762207007e-08,
+      "loss": 0.5519,
+      "step": 2259
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2197398447283404e-08,
+      "loss": 0.5469,
+      "step": 2260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1510493143782609e-08,
+      "loss": 0.5481,
+      "step": 2261
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0843483079755468e-08,
+      "loss": 0.5573,
+      "step": 2262
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0196369583763688e-08,
+      "loss": 0.5697,
+      "step": 2263
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.569153944740673e-09,
+      "loss": 0.5413,
+      "step": 2264
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.961837411982643e-09,
+      "loss": 0.5482,
+      "step": 2265
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.374421195155302e-09,
+      "loss": 0.5542,
+      "step": 2266
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.806906464281617e-09,
+      "loss": 0.5538,
+      "step": 2267
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.25929434974848e-09,
+      "loss": 0.5465,
+      "step": 2268
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.731585942297836e-09,
+      "loss": 0.5469,
+      "step": 2269
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.22378229302778e-09,
+      "loss": 0.5401,
+      "step": 2270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.735884413391457e-09,
+      "loss": 0.5419,
+      "step": 2271
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.267893275191505e-09,
+      "loss": 0.547,
+      "step": 2272
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.819809810578946e-09,
+      "loss": 0.5488,
+      "step": 2273
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.39163491205652e-09,
+      "loss": 0.5569,
+      "step": 2274
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9833694324686864e-09,
+      "loss": 0.5503,
+      "step": 2275
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5950141850038533e-09,
+      "loss": 0.5401,
+      "step": 2276
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.226569943197699e-09,
+      "loss": 0.561,
+      "step": 2277
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8780374409209667e-09,
+      "loss": 0.5513,
+      "step": 2278
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.549417372388341e-09,
+      "loss": 0.5523,
+      "step": 2279
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2407103921484595e-09,
+      "loss": 0.5456,
+      "step": 2280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.951917115091684e-09,
+      "loss": 0.5474,
+      "step": 2281
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6830381164389952e-09,
+      "loss": 0.5354,
+      "step": 2282
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4340739317497688e-09,
+      "loss": 0.5455,
+      "step": 2283
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.205025056914e-09,
+      "loss": 0.5485,
+      "step": 2284
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.958919481556362e-10,
+      "loss": 0.5581,
+      "step": 2285
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.066750220292463e-10,
+      "loss": 0.5431,
+      "step": 2286
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.3737465542002e-10,
+      "loss": 0.5582,
+      "step": 2287
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.879911855426578e-10,
+      "loss": 0.5338,
+      "step": 2288
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.585249099435917e-10,
+      "loss": 0.5517,
+      "step": 2289
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4897608649321335e-10,
+      "loss": 0.5599,
+      "step": 2290
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.593449333947561e-10,
+      "loss": 0.5431,
+      "step": 2291
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.963162917763335e-11,
+      "loss": 0.5428,
+      "step": 2292
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.983631269521837e-11,
+      "loss": 0.5455,
+      "step": 2293
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.959083132615733e-12,
+      "loss": 0.55,
+      "step": 2294
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.5474,
+      "step": 2295
+    },
+    {
+      "epoch": 3.0,
+      "step": 2295,
+      "total_flos": 6.080953256006124e+18,
+      "train_loss": 0.6370166003314498,
+      "train_runtime": 83683.912,
+      "train_samples_per_second": 3.514,
+      "train_steps_per_second": 0.027
+    }
+  ],
+  "max_steps": 2295,
+  "num_train_epochs": 3,
+  "total_flos": 6.080953256006124e+18,
+  "trial_name": null,
+  "trial_params": null
+}