diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4768 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9981078524124882,
+  "global_step": 792,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 1.1535,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 1.1146,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.1232,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.1125,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 1.1292,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1033,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 1.0263,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.0498,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.9317,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 0.9448,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 0.8364,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.8242,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 0.843,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 0.8247,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.7759,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.8293,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 0.8063,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.7742,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0555555555555557e-05,
+      "loss": 0.7717,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.7418,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.7428,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 0.7507,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 0.7293,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7476,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.7012,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 0.758,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.725,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.7158,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6111111111111115e-05,
+      "loss": 0.6808,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.725,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 0.7143,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.7062,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.7213,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.7042,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.6899,
+      "step": 35
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.7242,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999962815279418e-05,
+      "loss": 0.7249,
+      "step": 37
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999985126139422e-05,
+      "loss": 0.6757,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999665339174013e-05,
+      "loss": 0.6956,
+      "step": 39
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99994050500015e-05,
+      "loss": 0.7112,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999070395812432e-05,
+      "loss": 0.7057,
+      "step": 41
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998661379095622e-05,
+      "loss": 0.7118,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998178002892893e-05,
+      "loss": 0.7124,
+      "step": 43
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999762027079909e-05,
+      "loss": 0.7088,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996988186962044e-05,
+      "loss": 0.6741,
+      "step": 45
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999628175608252e-05,
+      "loss": 0.705,
+      "step": 46
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995500983414202e-05,
+      "loss": 0.6613,
+      "step": 47
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994645874763657e-05,
+      "loss": 0.6923,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993716436490283e-05,
+      "loss": 0.7003,
+      "step": 49
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992712675506253e-05,
+      "loss": 0.6868,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991634599276488e-05,
+      "loss": 0.6955,
+      "step": 51
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999048221581858e-05,
+      "loss": 0.6938,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989255533702736e-05,
+      "loss": 0.6769,
+      "step": 53
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987954562051724e-05,
+      "loss": 0.6775,
+      "step": 54
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99865793105408e-05,
+      "loss": 0.7066,
+      "step": 55
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985129789397633e-05,
+      "loss": 0.6729,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983606009402224e-05,
+      "loss": 0.7342,
+      "step": 57
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998200798188685e-05,
+      "loss": 0.6699,
+      "step": 58
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980335718735945e-05,
+      "loss": 0.6908,
+      "step": 59
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978589232386036e-05,
+      "loss": 0.7,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997676853582565e-05,
+      "loss": 0.6756,
+      "step": 61
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99748736425952e-05,
+      "loss": 0.6865,
+      "step": 62
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972904566786903e-05,
+      "loss": 0.7063,
+      "step": 63
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970861323044667e-05,
+      "loss": 0.6795,
+      "step": 64
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968743926563983e-05,
+      "loss": 0.7055,
+      "step": 65
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966552393091804e-05,
+      "loss": 0.6914,
+      "step": 66
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964286738926447e-05,
+      "loss": 0.6785,
+      "step": 67
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 0.682,
+      "step": 68
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995953313646548e-05,
+      "loss": 0.6519,
+      "step": 69
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995704522352214e-05,
+      "loss": 0.6804,
+      "step": 70
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954483260589914e-05,
+      "loss": 0.6871,
+      "step": 71
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995184726672197e-05,
+      "loss": 0.718,
+      "step": 72
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949137261522052e-05,
+      "loss": 0.662,
+      "step": 73
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946353265144315e-05,
+      "loss": 0.6987,
+      "step": 74
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943495298293182e-05,
+      "loss": 0.701,
+      "step": 75
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940563382223196e-05,
+      "loss": 0.6786,
+      "step": 76
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937557538738853e-05,
+      "loss": 0.6672,
+      "step": 77
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934477790194445e-05,
+      "loss": 0.665,
+      "step": 78
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931324159493884e-05,
+      "loss": 0.6163,
+      "step": 79
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928096670090552e-05,
+      "loss": 0.6811,
+      "step": 80
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924795345987103e-05,
+      "loss": 0.6944,
+      "step": 81
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99214202117353e-05,
+      "loss": 0.6697,
+      "step": 82
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917971292435827e-05,
+      "loss": 0.652,
+      "step": 83
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914448613738107e-05,
+      "loss": 0.6539,
+      "step": 84
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910852201840102e-05,
+      "loss": 0.6738,
+      "step": 85
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907182083488127e-05,
+      "loss": 0.6854,
+      "step": 86
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990343828597665e-05,
+      "loss": 0.6863,
+      "step": 87
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989962083714808e-05,
+      "loss": 0.7058,
+      "step": 88
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895729765392574e-05,
+      "loss": 0.6607,
+      "step": 89
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989176509964781e-05,
+      "loss": 0.6528,
+      "step": 90
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887726869398795e-05,
+      "loss": 0.6372,
+      "step": 91
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988361510467761e-05,
+      "loss": 0.6881,
+      "step": 92
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879429836063228e-05,
+      "loss": 0.6647,
+      "step": 93
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875171094681248e-05,
+      "loss": 0.6863,
+      "step": 94
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870838912203706e-05,
+      "loss": 0.6511,
+      "step": 95
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866433320848793e-05,
+      "loss": 0.6835,
+      "step": 96
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861954353380643e-05,
+      "loss": 0.6212,
+      "step": 97
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985740204310909e-05,
+      "loss": 0.6651,
+      "step": 98
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852776423889414e-05,
+      "loss": 0.6268,
+      "step": 99
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.6482,
+      "step": 100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843305396752507e-05,
+      "loss": 0.6927,
+      "step": 101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838460059270775e-05,
+      "loss": 0.6353,
+      "step": 102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833541553711395e-05,
+      "loss": 0.6431,
+      "step": 103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828549916653013e-05,
+      "loss": 0.6542,
+      "step": 104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982348518521816e-05,
+      "loss": 0.652,
+      "step": 105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818347397072954e-05,
+      "loss": 0.6993,
+      "step": 106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813136590426844e-05,
+      "loss": 0.6754,
+      "step": 107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 0.6415,
+      "step": 108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802496077184564e-05,
+      "loss": 0.7023,
+      "step": 109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797066449721295e-05,
+      "loss": 0.7072,
+      "step": 110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979156396202234e-05,
+      "loss": 0.6651,
+      "step": 111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785988655009386e-05,
+      "loss": 0.6827,
+      "step": 112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9780340570145683e-05,
+      "loss": 0.656,
+      "step": 113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977461974943572e-05,
+      "loss": 0.6091,
+      "step": 114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768826235424924e-05,
+      "loss": 0.6738,
+      "step": 115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762960071199334e-05,
+      "loss": 0.6478,
+      "step": 116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9757021300385288e-05,
+      "loss": 0.6687,
+      "step": 117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975100996714909e-05,
+      "loss": 0.6805,
+      "step": 118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744926116196687e-05,
+      "loss": 0.673,
+      "step": 119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738769792773338e-05,
+      "loss": 0.7099,
+      "step": 120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973254104266328e-05,
+      "loss": 0.6607,
+      "step": 121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726239912189382e-05,
+      "loss": 0.6523,
+      "step": 122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719866448212794e-05,
+      "loss": 0.7245,
+      "step": 123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713420698132614e-05,
+      "loss": 0.6925,
+      "step": 124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706902709885524e-05,
+      "loss": 0.6979,
+      "step": 125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700312531945444e-05,
+      "loss": 0.6214,
+      "step": 126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693650213323144e-05,
+      "loss": 0.6713,
+      "step": 127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686915803565934e-05,
+      "loss": 0.681,
+      "step": 128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9680109352757228e-05,
+      "loss": 0.6952,
+      "step": 129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967323091151623e-05,
+      "loss": 0.6709,
+      "step": 130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666280530997518e-05,
+      "loss": 0.6756,
+      "step": 131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.658,
+      "step": 132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652164159419945e-05,
+      "loss": 0.6245,
+      "step": 133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644998273343753e-05,
+      "loss": 0.6665,
+      "step": 134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96377606579544e-05,
+      "loss": 0.6493,
+      "step": 135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963045136707763e-05,
+      "loss": 0.6955,
+      "step": 136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623070455072225e-05,
+      "loss": 0.6439,
+      "step": 137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615617976829622e-05,
+      "loss": 0.6818,
+      "step": 138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960809398777348e-05,
+      "loss": 0.6422,
+      "step": 139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960049854385929e-05,
+      "loss": 0.7246,
+      "step": 140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959283170157394e-05,
+      "loss": 0.6474,
+      "step": 141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585093517935308e-05,
+      "loss": 0.6345,
+      "step": 142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577284050491833e-05,
+      "loss": 0.6896,
+      "step": 143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956940335732209e-05,
+      "loss": 0.6511,
+      "step": 144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561451497034354e-05,
+      "loss": 0.6599,
+      "step": 145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553428528766163e-05,
+      "loss": 0.675,
+      "step": 146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545334512183886e-05,
+      "loss": 0.6603,
+      "step": 147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953716950748227e-05,
+      "loss": 0.662,
+      "step": 148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528933575384e-05,
+      "loss": 0.6668,
+      "step": 149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520626777139243e-05,
+      "loss": 0.6663,
+      "step": 150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512249174525193e-05,
+      "loss": 0.6255,
+      "step": 151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503800829845613e-05,
+      "loss": 0.6124,
+      "step": 152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949528180593037e-05,
+      "loss": 0.6781,
+      "step": 153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486692166134964e-05,
+      "loss": 0.6726,
+      "step": 154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478031974340075e-05,
+      "loss": 0.6606,
+      "step": 155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946930129495106e-05,
+      "loss": 0.6452,
+      "step": 156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946050019289749e-05,
+      "loss": 0.604,
+      "step": 157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945162873363268e-05,
+      "loss": 0.646,
+      "step": 158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944268698313317e-05,
+      "loss": 0.6674,
+      "step": 159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433675007898255e-05,
+      "loss": 0.6607,
+      "step": 160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424592874949498e-05,
+      "loss": 0.6437,
+      "step": 161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941544065183021e-05,
+      "loss": 0.6249,
+      "step": 162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406218406604966e-05,
+      "loss": 0.67,
+      "step": 163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.6413,
+      "step": 164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387564124698133e-05,
+      "loss": 0.6238,
+      "step": 165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93781322267474e-05,
+      "loss": 0.6187,
+      "step": 166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936863058415138e-05,
+      "loss": 0.675,
+      "step": 167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935905926757326e-05,
+      "loss": 0.6016,
+      "step": 168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349418348194385e-05,
+      "loss": 0.6426,
+      "step": 169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339707897713737e-05,
+      "loss": 0.6358,
+      "step": 170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932992798834739e-05,
+      "loss": 0.6373,
+      "step": 171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932007869282799e-05,
+      "loss": 0.6645,
+      "step": 172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310160084404184e-05,
+      "loss": 0.6827,
+      "step": 173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930017223684012e-05,
+      "loss": 0.672,
+      "step": 174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290115224414862e-05,
+      "loss": 0.6683,
+      "step": 175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279989121921846e-05,
+      "loss": 0.6462,
+      "step": 176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926979400466833e-05,
+      "loss": 0.6812,
+      "step": 177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259529948474833e-05,
+      "loss": 0.6621,
+      "step": 178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249197029674566e-05,
+      "loss": 0.7096,
+      "step": 179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.665,
+      "step": 180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922832491214664e-05,
+      "loss": 0.6793,
+      "step": 181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921778586864375e-05,
+      "loss": 0.6602,
+      "step": 182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207178272982477e-05,
+      "loss": 0.6732,
+      "step": 183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196502204050925e-05,
+      "loss": 0.646,
+      "step": 184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918575774124641e-05,
+      "loss": 0.6659,
+      "step": 185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174944964474914e-05,
+      "loss": 0.6471,
+      "step": 186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164063954150445e-05,
+      "loss": 0.6523,
+      "step": 187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153114791194475e-05,
+      "loss": 0.686,
+      "step": 188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914209755703531e-05,
+      "loss": 0.6669,
+      "step": 189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131012333607507e-05,
+      "loss": 0.6569,
+      "step": 190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119859203351256e-05,
+      "loss": 0.6587,
+      "step": 191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910863824921176e-05,
+      "loss": 0.675,
+      "step": 192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097349554638632e-05,
+      "loss": 0.6495,
+      "step": 193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085993203585257e-05,
+      "loss": 0.6678,
+      "step": 194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074569280508188e-05,
+      "loss": 0.6778,
+      "step": 195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063077870366504e-05,
+      "loss": 0.6817,
+      "step": 196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905151905862117e-05,
+      "loss": 0.6854,
+      "step": 197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039892931234434e-05,
+      "loss": 0.6857,
+      "step": 198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028199574669153e-05,
+      "loss": 0.6424,
+      "step": 199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901643907588816e-05,
+      "loss": 0.6584,
+      "step": 200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900461152235364e-05,
+      "loss": 0.6161,
+      "step": 201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992717002026433e-05,
+      "loss": 0.7129,
+      "step": 202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898075560336543e-05,
+      "loss": 0.6358,
+      "step": 203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968727415326885e-05,
+      "loss": 0.6626,
+      "step": 204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956632527363755e-05,
+      "loss": 0.6885,
+      "step": 205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944471029425052e-05,
+      "loss": 0.6319,
+      "step": 206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932243011955154e-05,
+      "loss": 0.6501,
+      "step": 207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8919948565893144e-05,
+      "loss": 0.6706,
+      "step": 208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8907587782672127e-05,
+      "loss": 0.6525,
+      "step": 209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895160754218562e-05,
+      "loss": 0.6617,
+      "step": 210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882667572951562e-05,
+      "loss": 0.6079,
+      "step": 211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887010833178222e-05,
+      "loss": 0.6607,
+      "step": 212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885748312411291e-05,
+      "loss": 0.6087,
+      "step": 213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844792043836592e-05,
+      "loss": 0.6307,
+      "step": 214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832035185336122e-05,
+      "loss": 0.6839,
+      "step": 215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 0.6287,
+      "step": 216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88063245136394e-05,
+      "loss": 0.6377,
+      "step": 217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8793370891651973e-05,
+      "loss": 0.6755,
+      "step": 218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780351873856627e-05,
+      "loss": 0.6552,
+      "step": 219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876726755707508e-05,
+      "loss": 0.6464,
+      "step": 220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754118038614656e-05,
+      "loss": 0.6823,
+      "step": 221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874090341626759e-05,
+      "loss": 0.6111,
+      "step": 222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872762378831029e-05,
+      "loss": 0.6435,
+      "step": 223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8714279253502616e-05,
+      "loss": 0.6458,
+      "step": 224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8700869911087115e-05,
+      "loss": 0.6892,
+      "step": 225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687395860788325e-05,
+      "loss": 0.6426,
+      "step": 226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673857202812003e-05,
+      "loss": 0.647,
+      "step": 227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.6681,
+      "step": 228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646586467051464e-05,
+      "loss": 0.6382,
+      "step": 229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8632854592078185e-05,
+      "loss": 0.6696,
+      "step": 230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8619058515047746e-05,
+      "loss": 0.6322,
+      "step": 231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860519833856079e-05,
+      "loss": 0.6409,
+      "step": 232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591274165694686e-05,
+      "loss": 0.6464,
+      "step": 233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577286100002723e-05,
+      "loss": 0.6364,
+      "step": 234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563234245513366e-05,
+      "loss": 0.6958,
+      "step": 235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854911870672947e-05,
+      "loss": 0.655,
+      "step": 236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534939588627506e-05,
+      "loss": 0.67,
+      "step": 237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8520696996656787e-05,
+      "loss": 0.6337,
+      "step": 238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506391036738677e-05,
+      "loss": 0.6673,
+      "step": 239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849202181526579e-05,
+      "loss": 0.6569,
+      "step": 240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477589439101233e-05,
+      "loss": 0.6217,
+      "step": 241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8463094015577772e-05,
+      "loss": 0.6362,
+      "step": 242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448535652497073e-05,
+      "loss": 0.638,
+      "step": 243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843391445812886e-05,
+      "loss": 0.6515,
+      "step": 244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841923054121014e-05,
+      "loss": 0.678,
+      "step": 245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840448401094438e-05,
+      "loss": 0.6583,
+      "step": 246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838967497700071e-05,
+      "loss": 0.637,
+      "step": 247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837480354951308e-05,
+      "loss": 0.6371,
+      "step": 248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835986983907947e-05,
+      "loss": 0.6771,
+      "step": 249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344873956761045e-05,
+      "loss": 0.6367,
+      "step": 250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832981601408135e-05,
+      "loss": 0.6701,
+      "step": 251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 0.7181,
+      "step": 252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829951439603915e-05,
+      "loss": 0.6647,
+      "step": 253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8284270946028092e-05,
+      "loss": 0.6927,
+      "step": 254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268965886356965e-05,
+      "loss": 0.6299,
+      "step": 255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253599330848638e-05,
+      "loss": 0.6048,
+      "step": 256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238171393783335e-05,
+      "loss": 0.6462,
+      "step": 257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822268218989775e-05,
+      "loss": 0.6601,
+      "step": 258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820713183438424e-05,
+      "loss": 0.6416,
+      "step": 259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819152044288992e-05,
+      "loss": 0.6307,
+      "step": 260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817584813151584e-05,
+      "loss": 0.6809,
+      "step": 261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81601150168161e-05,
+      "loss": 0.6562,
+      "step": 262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144321215797e-05,
+      "loss": 0.678,
+      "step": 263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8128466845916156e-05,
+      "loss": 0.6343,
+      "step": 264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112552025081623e-05,
+      "loss": 0.662,
+      "step": 265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809657687165104e-05,
+      "loss": 0.6638,
+      "step": 266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8080541504430732e-05,
+      "loss": 0.6728,
+      "step": 267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806444604267483e-05,
+      "loss": 0.631,
+      "step": 268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8048290606084373e-05,
+      "loss": 0.6626,
+      "step": 269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803207531480645e-05,
+      "loss": 0.6873,
+      "step": 270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015800289433273e-05,
+      "loss": 0.6266,
+      "step": 271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999465651001297e-05,
+      "loss": 0.6404,
+      "step": 272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983071520990316e-05,
+      "loss": 0.6478,
+      "step": 273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966618021322558e-05,
+      "loss": 0.6623,
+      "step": 274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950105274361774e-05,
+      "loss": 0.6473,
+      "step": 275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933533402912354e-05,
+      "loss": 0.6641,
+      "step": 276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791690253021837e-05,
+      "loss": 0.6444,
+      "step": 277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790021277996269e-05,
+      "loss": 0.5904,
+      "step": 278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883464276266064e-05,
+      "loss": 0.679,
+      "step": 279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786665714368617e-05,
+      "loss": 0.6071,
+      "step": 280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784979150721672e-05,
+      "loss": 0.6554,
+      "step": 281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7832867492286506e-05,
+      "loss": 0.682,
+      "step": 282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781588522475848e-05,
+      "loss": 0.6609,
+      "step": 283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798844830928818e-05,
+      "loss": 0.6545,
+      "step": 284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778174643752598e-05,
+      "loss": 0.5987,
+      "step": 285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776459017170976e-05,
+      "loss": 0.6301,
+      "step": 286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7747376161070345e-05,
+      "loss": 0.6939,
+      "step": 287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773010453362737e-05,
+      "loss": 0.6788,
+      "step": 288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771277541782897e-05,
+      "loss": 0.6353,
+      "step": 289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7695388942550807e-05,
+      "loss": 0.696,
+      "step": 290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677945237095122e-05,
+      "loss": 0.6588,
+      "step": 291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.654,
+      "step": 292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76428866549873e-05,
+      "loss": 0.6903,
+      "step": 293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625272039063884e-05,
+      "loss": 0.6536,
+      "step": 294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607600714418437e-05,
+      "loss": 0.6744,
+      "step": 295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758987281247162e-05,
+      "loss": 0.6566,
+      "step": 296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572088465064847e-05,
+      "loss": 0.6868,
+      "step": 297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7554247804459317e-05,
+      "loss": 0.6592,
+      "step": 298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7536350963335023e-05,
+      "loss": 0.629,
+      "step": 299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7518398074789776e-05,
+      "loss": 0.6796,
+      "step": 300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7500389272338202e-05,
+      "loss": 0.6658,
+      "step": 301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748232468991076e-05,
+      "loss": 0.6874,
+      "step": 302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746420446185274e-05,
+      "loss": 0.6536,
+      "step": 303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7446028722923266e-05,
+      "loss": 0.6362,
+      "step": 304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427797608294286e-05,
+      "loss": 0.7043,
+      "step": 305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7409511253549592e-05,
+      "loss": 0.6955,
+      "step": 306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391169794683776e-05,
+      "loss": 0.6287,
+      "step": 307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737277336810124e-05,
+      "loss": 0.6316,
+      "step": 308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735432211061519e-05,
+      "loss": 0.6404,
+      "step": 309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7335816159446585e-05,
+      "loss": 0.5976,
+      "step": 310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7317255652223157e-05,
+      "loss": 0.6582,
+      "step": 311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298640726978357e-05,
+      "loss": 0.6285,
+      "step": 312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279971522150348e-05,
+      "loss": 0.6657,
+      "step": 313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726124817658096e-05,
+      "loss": 0.6436,
+      "step": 314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7242470829514674e-05,
+      "loss": 0.6437,
+      "step": 315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223639620597556e-05,
+      "loss": 0.6398,
+      "step": 316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720475468987627e-05,
+      "loss": 0.6327,
+      "step": 317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718581617779698e-05,
+      "loss": 0.6683,
+      "step": 318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716682422520435e-05,
+      "loss": 0.616,
+      "step": 319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147778973340466e-05,
+      "loss": 0.6649,
+      "step": 320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71286805638438e-05,
+      "loss": 0.6257,
+      "step": 321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109529138748156e-05,
+      "loss": 0.6568,
+      "step": 322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709032484048162e-05,
+      "loss": 0.6321,
+      "step": 323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.66,
+      "step": 324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051758196113172e-05,
+      "loss": 0.6597,
+      "step": 325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7032396136829247e-05,
+      "loss": 0.6481,
+      "step": 326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7012981778008245e-05,
+      "loss": 0.6527,
+      "step": 327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699351526403367e-05,
+      "loss": 0.6293,
+      "step": 328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6973996739676904e-05,
+      "loss": 0.6386,
+      "step": 329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6954426350096118e-05,
+      "loss": 0.6295,
+      "step": 330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934804240835204e-05,
+      "loss": 0.6577,
+      "step": 331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6915130557822698e-05,
+      "loss": 0.6057,
+      "step": 332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689540544737067e-05,
+      "loss": 0.6157,
+      "step": 333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6875629056173674e-05,
+      "loss": 0.6532,
+      "step": 334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855801531307625e-05,
+      "loss": 0.643,
+      "step": 335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6835923020228714e-05,
+      "loss": 0.6337,
+      "step": 336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681599367077232e-05,
+      "loss": 0.6768,
+      "step": 337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796013631151898e-05,
+      "loss": 0.6267,
+      "step": 338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6775983049957888e-05,
+      "loss": 0.6719,
+      "step": 339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6755902076156606e-05,
+      "loss": 0.6643,
+      "step": 340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6735770859089124e-05,
+      "loss": 0.6533,
+      "step": 341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715589548470187e-05,
+      "loss": 0.6686,
+      "step": 342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695358294387064e-05,
+      "loss": 0.6594,
+      "step": 343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6675077247298475e-05,
+      "loss": 0.6461,
+      "step": 344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6654746558033424e-05,
+      "loss": 0.6491,
+      "step": 345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6634366377790113e-05,
+      "loss": 0.6693,
+      "step": 346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6613936858134804e-05,
+      "loss": 0.6159,
+      "step": 347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659345815100069e-05,
+      "loss": 0.6579,
+      "step": 348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6572930408686777e-05,
+      "loss": 0.6522,
+      "step": 349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552353783856733e-05,
+      "loss": 0.66,
+      "step": 350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6531728429537766e-05,
+      "loss": 0.6473,
+      "step": 351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6511054499119493e-05,
+      "loss": 0.6593,
+      "step": 352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649033214635277e-05,
+      "loss": 0.6507,
+      "step": 353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469561525348576e-05,
+      "loss": 0.6432,
+      "step": 354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6448742790576862e-05,
+      "loss": 0.6542,
+      "step": 355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.615,
+      "step": 356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640696159939861e-05,
+      "loss": 0.635,
+      "step": 357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6385999453716453e-05,
+      "loss": 0.6493,
+      "step": 358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364989815713233e-05,
+      "loss": 0.6388,
+      "step": 359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343932841636455e-05,
+      "loss": 0.6945,
+      "step": 360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632282868808566e-05,
+      "loss": 0.6523,
+      "step": 361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6301677512011248e-05,
+      "loss": 0.6267,
+      "step": 362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6280479470713343e-05,
+      "loss": 0.6642,
+      "step": 363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6259234721840595e-05,
+      "loss": 0.651,
+      "step": 364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237943423389002e-05,
+      "loss": 0.6994,
+      "step": 365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6216605733700776e-05,
+      "loss": 0.6781,
+      "step": 366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195221811463114e-05,
+      "loss": 0.647,
+      "step": 367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6173791815707053e-05,
+      "loss": 0.6611,
+      "step": 368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615231590580627e-05,
+      "loss": 0.6056,
+      "step": 369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6130794241475912e-05,
+      "loss": 0.6426,
+      "step": 370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6109226982771393e-05,
+      "loss": 0.6162,
+      "step": 371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608761429008721e-05,
+      "loss": 0.6567,
+      "step": 372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065956324155746e-05,
+      "loss": 0.6676,
+      "step": 373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604425324604609e-05,
+      "loss": 0.6566,
+      "step": 374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6022505217162824e-05,
+      "loss": 0.6384,
+      "step": 375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6000712399244813e-05,
+      "loss": 0.6585,
+      "step": 376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5978874954364032e-05,
+      "loss": 0.6851,
+      "step": 377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5956993044924334e-05,
+      "loss": 0.6876,
+      "step": 378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5935066833660262e-05,
+      "loss": 0.6691,
+      "step": 379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5913096483635827e-05,
+      "loss": 0.6026,
+      "step": 380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5891082158243283e-05,
+      "loss": 0.6354,
+      "step": 381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586902402120195e-05,
+      "loss": 0.6356,
+      "step": 382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584692223655695e-05,
+      "loss": 0.6428,
+      "step": 383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5824776968678024e-05,
+      "loss": 0.6567,
+      "step": 384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5802588382258292e-05,
+      "loss": 0.5974,
+      "step": 385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5780356642313034e-05,
+      "loss": 0.6577,
+      "step": 386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5758081914178457e-05,
+      "loss": 0.6272,
+      "step": 387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573576436351046e-05,
+      "loss": 0.6461,
+      "step": 388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5713404156283435e-05,
+      "loss": 0.6271,
+      "step": 389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691001458788984e-05,
+      "loss": 0.6422,
+      "step": 390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5668556437634718e-05,
+      "loss": 0.6538,
+      "step": 391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5646069259743007e-05,
+      "loss": 0.6347,
+      "step": 392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5623540092349735e-05,
+      "loss": 0.6206,
+      "step": 393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5600969103003056e-05,
+      "loss": 0.6796,
+      "step": 394
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5578356459562163e-05,
+      "loss": 0.693,
+      "step": 395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.6679,
+      "step": 396
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533006883382124e-05,
+      "loss": 0.6109,
+      "step": 397
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5510270287905243e-05,
+      "loss": 0.536,
+      "step": 398
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548749271285616e-05,
+      "loss": 0.5578,
+      "step": 399
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5464674327630437e-05,
+      "loss": 0.5173,
+      "step": 400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544181530192712e-05,
+      "loss": 0.5399,
+      "step": 401
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5418915805747518e-05,
+      "loss": 0.5078,
+      "step": 402
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5395976009393898e-05,
+      "loss": 0.5134,
+      "step": 403
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5372996083468242e-05,
+      "loss": 0.5183,
+      "step": 404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349976198870974e-05,
+      "loss": 0.4931,
+      "step": 405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532691652679969e-05,
+      "loss": 0.5068,
+      "step": 406
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5303817238747892e-05,
+      "loss": 0.5361,
+      "step": 407
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528067850650368e-05,
+      "loss": 0.497,
+      "step": 408
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257500502148523e-05,
+      "loss": 0.5212,
+      "step": 409
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523428339805594e-05,
+      "loss": 0.5419,
+      "step": 410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211027366890237e-05,
+      "loss": 0.5199,
+      "step": 411
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187732581605217e-05,
+      "loss": 0.5118,
+      "step": 412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164399215442897e-05,
+      "loss": 0.5158,
+      "step": 413
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5141027441932217e-05,
+      "loss": 0.5401,
+      "step": 414
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5117617434887754e-05,
+      "loss": 0.5485,
+      "step": 415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509416936840842e-05,
+      "loss": 0.5242,
+      "step": 416
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5070683416876172e-05,
+      "loss": 0.5253,
+      "step": 417
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5047159754954721e-05,
+      "loss": 0.5049,
+      "step": 418
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502359855758823e-05,
+      "loss": 0.5089,
+      "step": 419
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5241,
+      "step": 420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4976364257691191e-05,
+      "loss": 0.5158,
+      "step": 421
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4952691506439497e-05,
+      "loss": 0.4904,
+      "step": 422
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928981922297842e-05,
+      "loss": 0.5496,
+      "step": 423
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905235681593079e-05,
+      "loss": 0.5279,
+      "step": 424
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4881452960924679e-05,
+      "loss": 0.5221,
+      "step": 425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4857633937163402e-05,
+      "loss": 0.516,
+      "step": 426
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833778787449997e-05,
+      "loss": 0.5135,
+      "step": 427
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4809887689193878e-05,
+      "loss": 0.5119,
+      "step": 428
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785960820071812e-05,
+      "loss": 0.5554,
+      "step": 429
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4761998358026581e-05,
+      "loss": 0.52,
+      "step": 430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473800048126568e-05,
+      "loss": 0.5489,
+      "step": 431
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713967368259981e-05,
+      "loss": 0.5249,
+      "step": 432
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689899197742391e-05,
+      "loss": 0.4994,
+      "step": 433
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4665796148706561e-05,
+      "loss": 0.5416,
+      "step": 434
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4641658400405516e-05,
+      "loss": 0.5295,
+      "step": 435
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617486132350343e-05,
+      "loss": 0.4802,
+      "step": 436
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4593279524308841e-05,
+      "loss": 0.4715,
+      "step": 437
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4569038756304209e-05,
+      "loss": 0.5009,
+      "step": 438
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544764008613681e-05,
+      "loss": 0.5346,
+      "step": 439
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45204554617672e-05,
+      "loss": 0.5295,
+      "step": 440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496113296546068e-05,
+      "loss": 0.5169,
+      "step": 441
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.447173769398161e-05,
+      "loss": 0.5137,
+      "step": 442
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444732883535382e-05,
+      "loss": 0.5308,
+      "step": 443
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4422886902190014e-05,
+      "loss": 0.5483,
+      "step": 444
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4398412076263486e-05,
+      "loss": 0.5392,
+      "step": 445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373904539592145e-05,
+      "loss": 0.5411,
+      "step": 446
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4349364474437172e-05,
+      "loss": 0.5076,
+      "step": 447
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324792063301662e-05,
+      "loss": 0.5131,
+      "step": 448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4300187488929264e-05,
+      "loss": 0.5248,
+      "step": 449
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4275550934302822e-05,
+      "loss": 0.5201,
+      "step": 450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4250882582643012e-05,
+      "loss": 0.4962,
+      "step": 451
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4226182617406996e-05,
+      "loss": 0.5044,
+      "step": 452
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4201451222287026e-05,
+      "loss": 0.5114,
+      "step": 453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4176688581209109e-05,
+      "loss": 0.5494,
+      "step": 454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4151894878331622e-05,
+      "loss": 0.536,
+      "step": 455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4127070298043949e-05,
+      "loss": 0.552,
+      "step": 456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102215024965107e-05,
+      "loss": 0.5262,
+      "step": 457
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077329243942368e-05,
+      "loss": 0.5205,
+      "step": 458
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4052413140049898e-05,
+      "loss": 0.511,
+      "step": 459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4027466898587375e-05,
+      "loss": 0.5229,
+      "step": 460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4002490705078593e-05,
+      "loss": 0.5125,
+      "step": 461
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3977484745270112e-05,
+      "loss": 0.5448,
+      "step": 462
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3952449205129856e-05,
+      "loss": 0.5387,
+      "step": 463
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927384270845744e-05,
+      "loss": 0.5316,
+      "step": 464
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902290128824282e-05,
+      "loss": 0.5323,
+      "step": 465
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3877166965689206e-05,
+      "loss": 0.5288,
+      "step": 466
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852014968280081e-05,
+      "loss": 0.5032,
+      "step": 467
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.5729,
+      "step": 468
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380162521906871e-05,
+      "loss": 0.5321,
+      "step": 469
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776387842012217e-05,
+      "loss": 0.518,
+      "step": 470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375112238017038e-05,
+      "loss": 0.5223,
+      "step": 471
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372582902144103e-05,
+      "loss": 0.5465,
+      "step": 472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700507953929464e-05,
+      "loss": 0.499,
+      "step": 473
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675159365947038e-05,
+      "loss": 0.4855,
+      "step": 474
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364978344600979e-05,
+      "loss": 0.5307,
+      "step": 475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624380382837017e-05,
+      "loss": 0.5115,
+      "step": 476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3598950365349884e-05,
+      "loss": 0.5412,
+      "step": 477
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573493582670003e-05,
+      "loss": 0.5442,
+      "step": 478
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354801022411805e-05,
+      "loss": 0.5576,
+      "step": 479
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3522500479212337e-05,
+      "loss": 0.5183,
+      "step": 480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349696453766741e-05,
+      "loss": 0.5001,
+      "step": 481
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347140258939264e-05,
+      "loss": 0.5122,
+      "step": 482
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3445814824490806e-05,
+      "loss": 0.5509,
+      "step": 483
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.5429,
+      "step": 484
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3394562606175647e-05,
+      "loss": 0.5083,
+      "step": 485
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3368898533922202e-05,
+      "loss": 0.5192,
+      "step": 486
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3343209407358628e-05,
+      "loss": 0.5184,
+      "step": 487
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3317495417533523e-05,
+      "loss": 0.5099,
+      "step": 488
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.329175675568039e-05,
+      "loss": 0.5153,
+      "step": 489
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3265993613216223e-05,
+      "loss": 0.5094,
+      "step": 490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240206181740069e-05,
+      "loss": 0.5324,
+      "step": 491
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3214394653031616e-05,
+      "loss": 0.5234,
+      "step": 492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3188559219049761e-05,
+      "loss": 0.5161,
+      "step": 493
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162700071931185e-05,
+      "loss": 0.4995,
+      "step": 494
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136817403988918e-05,
+      "loss": 0.5365,
+      "step": 495
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3110911407710909e-05,
+      "loss": 0.5392,
+      "step": 496
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3084982275758615e-05,
+      "loss": 0.5494,
+      "step": 497
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059030200965536e-05,
+      "loss": 0.537,
+      "step": 498
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033055376335804e-05,
+      "loss": 0.532,
+      "step": 499
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300705799504273e-05,
+      "loss": 0.5308,
+      "step": 500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981038250427402e-05,
+      "loss": 0.5444,
+      "step": 501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295499633599719e-05,
+      "loss": 0.5094,
+      "step": 502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2928932445424368e-05,
+      "loss": 0.527,
+      "step": 503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2902846772544625e-05,
+      "loss": 0.5301,
+      "step": 504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876739511355659e-05,
+      "loss": 0.5289,
+      "step": 505
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285061085601571e-05,
+      "loss": 0.534,
+      "step": 506
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824461000842126e-05,
+      "loss": 0.5212,
+      "step": 507
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798290140309924e-05,
+      "loss": 0.5618,
+      "step": 508
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2772098469050323e-05,
+      "loss": 0.5227,
+      "step": 509
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2745886181849325e-05,
+      "loss": 0.5452,
+      "step": 510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719653473646243e-05,
+      "loss": 0.5229,
+      "step": 511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2693400539532263e-05,
+      "loss": 0.4945,
+      "step": 512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2667127574748985e-05,
+      "loss": 0.5168,
+      "step": 513
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2640834774686985e-05,
+      "loss": 0.5469,
+      "step": 514
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2614522334884345e-05,
+      "loss": 0.5444,
+      "step": 515
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.5208,
+      "step": 516
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2561839318938327e-05,
+      "loss": 0.5474,
+      "step": 517
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2535469134595598e-05,
+      "loss": 0.5222,
+      "step": 518
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2509080094110604e-05,
+      "loss": 0.5253,
+      "step": 519
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2482672393737164e-05,
+      "loss": 0.5298,
+      "step": 520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2456246229867876e-05,
+      "loss": 0.5148,
+      "step": 521
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.242980179903264e-05,
+      "loss": 0.5504,
+      "step": 522
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240333929789721e-05,
+      "loss": 0.497,
+      "step": 523
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376858923261732e-05,
+      "loss": 0.5758,
+      "step": 524
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2350360872059268e-05,
+      "loss": 0.5359,
+      "step": 525
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2323845341354347e-05,
+      "loss": 0.5115,
+      "step": 526
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2297312528341484e-05,
+      "loss": 0.5455,
+      "step": 527
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2270762630343734e-05,
+      "loss": 0.5011,
+      "step": 528
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.22441958448112e-05,
+      "loss": 0.5141,
+      "step": 529
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221761236931958e-05,
+      "loss": 0.5023,
+      "step": 530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191012401568698e-05,
+      "loss": 0.5152,
+      "step": 531
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164396139381029e-05,
+      "loss": 0.4982,
+      "step": 532
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2137763780700226e-05,
+      "loss": 0.5374,
+      "step": 533
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2111115523589651e-05,
+      "loss": 0.5419,
+      "step": 534
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2084451566230905e-05,
+      "loss": 0.5419,
+      "step": 535
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205777210692235e-05,
+      "loss": 0.5631,
+      "step": 536
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.203107734407763e-05,
+      "loss": 0.5313,
+      "step": 537
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004367476224206e-05,
+      "loss": 0.4955,
+      "step": 538
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1977642702001874e-05,
+      "loss": 0.5361,
+      "step": 539
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.5176,
+      "step": 540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.192414922956247e-05,
+      "loss": 0.5387,
+      "step": 541
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1897380929173365e-05,
+      "loss": 0.529,
+      "step": 542
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1870598518068321e-05,
+      "loss": 0.5457,
+      "step": 543
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1843802195426634e-05,
+      "loss": 0.5263,
+      "step": 544
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1816992160531062e-05,
+      "loss": 0.544,
+      "step": 545
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1790168612766331e-05,
+      "loss": 0.5054,
+      "step": 546
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1763331751617663e-05,
+      "loss": 0.5181,
+      "step": 547
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.5895,
+      "step": 548
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1709618887603013e-05,
+      "loss": 0.5457,
+      "step": 549
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1682743284196595e-05,
+      "loss": 0.5179,
+      "step": 550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165585516632241e-05,
+      "loss": 0.5363,
+      "step": 551
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162895473394589e-05,
+      "loss": 0.5373,
+      "step": 552
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602042187124045e-05,
+      "loss": 0.5182,
+      "step": 553
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1575117726003979e-05,
+      "loss": 0.4936,
+      "step": 554
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1548181550821411e-05,
+      "loss": 0.5397,
+      "step": 555
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1521233861899168e-05,
+      "loss": 0.521,
+      "step": 556
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1494274859645707e-05,
+      "loss": 0.5062,
+      "step": 557
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467304744553618e-05,
+      "loss": 0.5098,
+      "step": 558
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1440323717198145e-05,
+      "loss": 0.5391,
+      "step": 559
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413331978235677e-05,
+      "loss": 0.5311,
+      "step": 560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386329728402269e-05,
+      "loss": 0.5341,
+      "step": 561
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1359317168512143e-05,
+      "loss": 0.5275,
+      "step": 562
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13322944994562e-05,
+      "loss": 0.4977,
+      "step": 563
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130526192220052e-05,
+      "loss": 0.5518,
+      "step": 564
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1278219637784862e-05,
+      "loss": 0.5128,
+      "step": 565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1251167847321194e-05,
+      "loss": 0.5059,
+      "step": 566
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224106751992164e-05,
+      "loss": 0.5233,
+      "step": 567
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1197036553049626e-05,
+      "loss": 0.5472,
+      "step": 568
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1169957451813138e-05,
+      "loss": 0.5021,
+      "step": 569
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1142869649668467e-05,
+      "loss": 0.522,
+      "step": 570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1115773348066074e-05,
+      "loss": 0.5125,
+      "step": 571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1088668748519646e-05,
+      "loss": 0.5035,
+      "step": 572
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1061556052604579e-05,
+      "loss": 0.5425,
+      "step": 573
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034435461956465e-05,
+      "loss": 0.4998,
+      "step": 574
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1007307178269633e-05,
+      "loss": 0.4947,
+      "step": 575
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 0.5462,
+      "step": 576
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095302833884163e-05,
+      "loss": 0.5321,
+      "step": 577
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0925878186769159e-05,
+      "loss": 0.5485,
+      "step": 578
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089872114899235e-05,
+      "loss": 0.5064,
+      "step": 579
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0871557427476585e-05,
+      "loss": 0.535,
+      "step": 580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0844387224236931e-05,
+      "loss": 0.5478,
+      "step": 581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0817210741336684e-05,
+      "loss": 0.5351,
+      "step": 582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079002818088582e-05,
+      "loss": 0.5355,
+      "step": 583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0762839745039526e-05,
+      "loss": 0.5796,
+      "step": 584
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735645635996676e-05,
+      "loss": 0.4861,
+      "step": 585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0708446055998342e-05,
+      "loss": 0.5507,
+      "step": 586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681241207326283e-05,
+      "loss": 0.5487,
+      "step": 587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0654031292301432e-05,
+      "loss": 0.5354,
+      "step": 588
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0626816513282412e-05,
+      "loss": 0.5376,
+      "step": 589
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0599597072664012e-05,
+      "loss": 0.5234,
+      "step": 590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057237317287569e-05,
+      "loss": 0.5293,
+      "step": 591
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545145016380065e-05,
+      "loss": 0.5383,
+      "step": 592
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051791280567142e-05,
+      "loss": 0.5256,
+      "step": 593
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0490676743274181e-05,
+      "loss": 0.5127,
+      "step": 594
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0463437031741423e-05,
+      "loss": 0.5001,
+      "step": 595
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0436193873653362e-05,
+      "loss": 0.546,
+      "step": 596
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408947471615836e-05,
+      "loss": 0.5442,
+      "step": 597
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0381698028258817e-05,
+      "loss": 0.536,
+      "step": 598
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0354445746234893e-05,
+      "loss": 0.5211,
+      "step": 599
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0327190828217763e-05,
+      "loss": 0.5769,
+      "step": 600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0299933476900728e-05,
+      "loss": 0.5344,
+      "step": 601
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272673894995187e-05,
+      "loss": 0.5108,
+      "step": 602
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0245412285229124e-05,
+      "loss": 0.4932,
+      "step": 603
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0218148850345613e-05,
+      "loss": 0.5415,
+      "step": 604
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0190883793101292e-05,
+      "loss": 0.5262,
+      "step": 605
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0163617316264869e-05,
+      "loss": 0.5268,
+      "step": 606
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0136349622615609e-05,
+      "loss": 0.5522,
+      "step": 607
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0109080914941825e-05,
+      "loss": 0.5352,
+      "step": 608
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0081811396039374e-05,
+      "loss": 0.5553,
+      "step": 609
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0054541268710139e-05,
+      "loss": 0.524,
+      "step": 610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002727073576054e-05,
+      "loss": 0.5283,
+      "step": 611
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.5253,
+      "step": 612
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972729264239461e-06,
+      "loss": 0.5006,
+      "step": 613
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.945458731289863e-06,
+      "loss": 0.5509,
+      "step": 614
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918188603960632e-06,
+      "loss": 0.536,
+      "step": 615
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.890919085058179e-06,
+      "loss": 0.5479,
+      "step": 616
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863650377384395e-06,
+      "loss": 0.5371,
+      "step": 617
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.836382683735133e-06,
+      "loss": 0.5169,
+      "step": 618
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.809116206898712e-06,
+      "loss": 0.5632,
+      "step": 619
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.78185114965439e-06,
+      "loss": 0.5008,
+      "step": 620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75458771477088e-06,
+      "loss": 0.5523,
+      "step": 621
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.727326105004818e-06,
+      "loss": 0.535,
+      "step": 622
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.700066523099273e-06,
+      "loss": 0.4995,
+      "step": 623
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.67280917178224e-06,
+      "loss": 0.5314,
+      "step": 624
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64555425376511e-06,
+      "loss": 0.539,
+      "step": 625
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.618301971741185e-06,
+      "loss": 0.507,
+      "step": 626
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591052528384168e-06,
+      "loss": 0.5244,
+      "step": 627
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.563806126346643e-06,
+      "loss": 0.5695,
+      "step": 628
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536562968258579e-06,
+      "loss": 0.5377,
+      "step": 629
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50932325672582e-06,
+      "loss": 0.5406,
+      "step": 630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.482087194328584e-06,
+      "loss": 0.5577,
+      "step": 631
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.454854983619936e-06,
+      "loss": 0.5282,
+      "step": 632
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427626827124316e-06,
+      "loss": 0.5268,
+      "step": 633
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.400402927335992e-06,
+      "loss": 0.5318,
+      "step": 634
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.373183486717592e-06,
+      "loss": 0.542,
+      "step": 635
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.34596870769857e-06,
+      "loss": 0.5313,
+      "step": 636
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.318758792673722e-06,
+      "loss": 0.5582,
+      "step": 637
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29155394400166e-06,
+      "loss": 0.5125,
+      "step": 638
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.264354364003327e-06,
+      "loss": 0.5039,
+      "step": 639
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.237160254960477e-06,
+      "loss": 0.5421,
+      "step": 640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209971819114181e-06,
+      "loss": 0.4873,
+      "step": 641
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182789258663321e-06,
+      "loss": 0.5128,
+      "step": 642
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15561277576307e-06,
+      "loss": 0.5089,
+      "step": 643
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.128442572523418e-06,
+      "loss": 0.5269,
+      "step": 644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.101278851007651e-06,
+      "loss": 0.5662,
+      "step": 645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.074121813230846e-06,
+      "loss": 0.5187,
+      "step": 646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.046971661158373e-06,
+      "loss": 0.5475,
+      "step": 647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.019828596704394e-06,
+      "loss": 0.5488,
+      "step": 648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.992692821730368e-06,
+      "loss": 0.5129,
+      "step": 649
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.965564538043535e-06,
+      "loss": 0.5544,
+      "step": 650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.938443947395428e-06,
+      "loss": 0.5401,
+      "step": 651
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.911331251480357e-06,
+      "loss": 0.5257,
+      "step": 652
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.884226651933929e-06,
+      "loss": 0.5299,
+      "step": 653
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.857130350331535e-06,
+      "loss": 0.5062,
+      "step": 654
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.830042548186864e-06,
+      "loss": 0.5039,
+      "step": 655
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.802963446950378e-06,
+      "loss": 0.5083,
+      "step": 656
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77589324800784e-06,
+      "loss": 0.5232,
+      "step": 657
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74883215267881e-06,
+      "loss": 0.5056,
+      "step": 658
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72178036221514e-06,
+      "loss": 0.532,
+      "step": 659
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.694738077799487e-06,
+      "loss": 0.511,
+      "step": 660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.667705500543802e-06,
+      "loss": 0.5245,
+      "step": 661
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.64068283148786e-06,
+      "loss": 0.5222,
+      "step": 662
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613670271597733e-06,
+      "loss": 0.5243,
+      "step": 663
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586668021764328e-06,
+      "loss": 0.5,
+      "step": 664
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55967628280186e-06,
+      "loss": 0.5179,
+      "step": 665
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532695255446384e-06,
+      "loss": 0.5303,
+      "step": 666
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.505725140354295e-06,
+      "loss": 0.5049,
+      "step": 667
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.478766138100834e-06,
+      "loss": 0.5377,
+      "step": 668
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.451818449178592e-06,
+      "loss": 0.5555,
+      "step": 669
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.424882273996023e-06,
+      "loss": 0.5067,
+      "step": 670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397957812875958e-06,
+      "loss": 0.4929,
+      "step": 671
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.371045266054114e-06,
+      "loss": 0.5521,
+      "step": 672
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.344144833677595e-06,
+      "loss": 0.4817,
+      "step": 673
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.317256715803407e-06,
+      "loss": 0.542,
+      "step": 674
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.290381112396989e-06,
+      "loss": 0.5292,
+      "step": 675
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.5447,
+      "step": 676
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.236668248382337e-06,
+      "loss": 0.5259,
+      "step": 677
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209831387233675e-06,
+      "loss": 0.5331,
+      "step": 678
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.183007839468942e-06,
+      "loss": 0.5282,
+      "step": 679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.156197804573368e-06,
+      "loss": 0.5025,
+      "step": 680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12940148193168e-06,
+      "loss": 0.5348,
+      "step": 681
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10261907082664e-06,
+      "loss": 0.5611,
+      "step": 682
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.075850770437535e-06,
+      "loss": 0.5139,
+      "step": 683
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.5122,
+      "step": 684
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.022357297998128e-06,
+      "loss": 0.5584,
+      "step": 685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995632523775795e-06,
+      "loss": 0.5301,
+      "step": 686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.968922655922375e-06,
+      "loss": 0.5388,
+      "step": 687
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.942227893077652e-06,
+      "loss": 0.5565,
+      "step": 688
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.915548433769096e-06,
+      "loss": 0.4994,
+      "step": 689
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888884476410348e-06,
+      "loss": 0.4997,
+      "step": 690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.862236219299777e-06,
+      "loss": 0.5531,
+      "step": 691
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835603860618973e-06,
+      "loss": 0.5599,
+      "step": 692
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808987598431303e-06,
+      "loss": 0.5255,
+      "step": 693
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.782387630680422e-06,
+      "loss": 0.4906,
+      "step": 694
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.755804155188801e-06,
+      "loss": 0.5125,
+      "step": 695
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72923736965627e-06,
+      "loss": 0.5271,
+      "step": 696
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702687471658519e-06,
+      "loss": 0.516,
+      "step": 697
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676154658645656e-06,
+      "loss": 0.5685,
+      "step": 698
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.649639127940734e-06,
+      "loss": 0.5114,
+      "step": 699
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.623141076738271e-06,
+      "loss": 0.4933,
+      "step": 700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.596660702102791e-06,
+      "loss": 0.5362,
+      "step": 701
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.570198200967363e-06,
+      "loss": 0.5292,
+      "step": 702
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.543753770132127e-06,
+      "loss": 0.5256,
+      "step": 703
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5173276062628364e-06,
+      "loss": 0.5509,
+      "step": 704
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.490919905889403e-06,
+      "loss": 0.5293,
+      "step": 705
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4645308654044065e-06,
+      "loss": 0.5245,
+      "step": 706
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.438160681061676e-06,
+      "loss": 0.5463,
+      "step": 707
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.5194,
+      "step": 708
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.385477665115659e-06,
+      "loss": 0.5365,
+      "step": 709
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.359165225313019e-06,
+      "loss": 0.5315,
+      "step": 710
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.332872425251017e-06,
+      "loss": 0.5268,
+      "step": 711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.306599460467741e-06,
+      "loss": 0.5158,
+      "step": 712
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.280346526353759e-06,
+      "loss": 0.5018,
+      "step": 713
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.25411381815068e-06,
+      "loss": 0.5108,
+      "step": 714
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.22790153094968e-06,
+      "loss": 0.5425,
+      "step": 715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.201709859690081e-06,
+      "loss": 0.5326,
+      "step": 716
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.175538999157876e-06,
+      "loss": 0.5131,
+      "step": 717
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.149389143984295e-06,
+      "loss": 0.5305,
+      "step": 718
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.123260488644345e-06,
+      "loss": 0.5096,
+      "step": 719
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097153227455379e-06,
+      "loss": 0.5146,
+      "step": 720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.071067554575637e-06,
+      "loss": 0.527,
+      "step": 721
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.04500366400281e-06,
+      "loss": 0.5189,
+      "step": 722
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.018961749572604e-06,
+      "loss": 0.5246,
+      "step": 723
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.992942004957271e-06,
+      "loss": 0.4796,
+      "step": 724
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9669446236642e-06,
+      "loss": 0.5277,
+      "step": 725
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940969799034465e-06,
+      "loss": 0.5042,
+      "step": 726
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9150177242413885e-06,
+      "loss": 0.5275,
+      "step": 727
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.889088592289092e-06,
+      "loss": 0.516,
+      "step": 728
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8631825960110866e-06,
+      "loss": 0.5046,
+      "step": 729
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8372999280688175e-06,
+      "loss": 0.5314,
+      "step": 730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.811440780950238e-06,
+      "loss": 0.5155,
+      "step": 731
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.785605346968387e-06,
+      "loss": 0.5146,
+      "step": 732
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7597938182599335e-06,
+      "loss": 0.5544,
+      "step": 733
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.73400638678378e-06,
+      "loss": 0.5456,
+      "step": 734
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.708243244319611e-06,
+      "loss": 0.5269,
+      "step": 735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.682504582466482e-06,
+      "loss": 0.538,
+      "step": 736
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.656790592641376e-06,
+      "loss": 0.5231,
+      "step": 737
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.631101466077801e-06,
+      "loss": 0.5297,
+      "step": 738
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.605437393824355e-06,
+      "loss": 0.4868,
+      "step": 739
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.5128,
+      "step": 740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.554185175509196e-06,
+      "loss": 0.5192,
+      "step": 741
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.528597410607364e-06,
+      "loss": 0.5038,
+      "step": 742
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.503035462332592e-06,
+      "loss": 0.5213,
+      "step": 743
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4774995207876654e-06,
+      "loss": 0.5272,
+      "step": 744
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451989775881955e-06,
+      "loss": 0.5195,
+      "step": 745
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.42650641733e-06,
+      "loss": 0.5221,
+      "step": 746
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.401049634650119e-06,
+      "loss": 0.5402,
+      "step": 747
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.375619617162985e-06,
+      "loss": 0.5196,
+      "step": 748
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.350216553990212e-06,
+      "loss": 0.5155,
+      "step": 749
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3248406340529665e-06,
+      "loss": 0.5284,
+      "step": 750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.2994920460705415e-06,
+      "loss": 0.5309,
+      "step": 751
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274170978558971e-06,
+      "loss": 0.5146,
+      "step": 752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2488776198296195e-06,
+      "loss": 0.5445,
+      "step": 753
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.223612157987786e-06,
+      "loss": 0.5109,
+      "step": 754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.1983747809312935e-06,
+      "loss": 0.5296,
+      "step": 755
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.5242,
+      "step": 756
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147985031719921e-06,
+      "loss": 0.5304,
+      "step": 757
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.122833034310794e-06,
+      "loss": 0.5278,
+      "step": 758
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.097709871175723e-06,
+      "loss": 0.5289,
+      "step": 759
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.072615729154261e-06,
+      "loss": 0.5122,
+      "step": 760
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0475507948701454e-06,
+      "loss": 0.5383,
+      "step": 761
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.02251525472989e-06,
+      "loss": 0.5242,
+      "step": 762
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.997509294921412e-06,
+      "loss": 0.5025,
+      "step": 763
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.97253310141263e-06,
+      "loss": 0.5159,
+      "step": 764
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947586859950103e-06,
+      "loss": 0.5558,
+      "step": 765
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.922670756057633e-06,
+      "loss": 0.505,
+      "step": 766
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.897784975034895e-06,
+      "loss": 0.5232,
+      "step": 767
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.872929701956054e-06,
+      "loss": 0.5281,
+      "step": 768
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.848105121668381e-06,
+      "loss": 0.5184,
+      "step": 769
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.823311418790894e-06,
+      "loss": 0.5378,
+      "step": 770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.798548777712977e-06,
+      "loss": 0.5432,
+      "step": 771
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.773817382593008e-06,
+      "loss": 0.5138,
+      "step": 772
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7491174173569885e-06,
+      "loss": 0.5127,
+      "step": 773
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.724449065697182e-06,
+      "loss": 0.5348,
+      "step": 774
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.699812511070735e-06,
+      "loss": 0.5341,
+      "step": 775
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.675207936698337e-06,
+      "loss": 0.5301,
+      "step": 776
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.650635525562829e-06,
+      "loss": 0.5176,
+      "step": 777
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6260954604078585e-06,
+      "loss": 0.5451,
+      "step": 778
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.601587923736518e-06,
+      "loss": 0.5305,
+      "step": 779
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5771130978099896e-06,
+      "loss": 0.543,
+      "step": 780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552671164646184e-06,
+      "loss": 0.5224,
+      "step": 781
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5282623060183945e-06,
+      "loss": 0.5455,
+      "step": 782
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.503886703453933e-06,
+      "loss": 0.5485,
+      "step": 783
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.479544538232804e-06,
+      "loss": 0.5384,
+      "step": 784
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455235991386322e-06,
+      "loss": 0.5385,
+      "step": 785
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.430961243695794e-06,
+      "loss": 0.5263,
+      "step": 786
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.406720475691163e-06,
+      "loss": 0.5396,
+      "step": 787
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382513867649663e-06,
+      "loss": 0.5423,
+      "step": 788
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.358341599594483e-06,
+      "loss": 0.5075,
+      "step": 789
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.334203851293442e-06,
+      "loss": 0.537,
+      "step": 790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.310100802257613e-06,
+      "loss": 0.5148,
+      "step": 791
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286032631740023e-06,
+      "loss": 0.5262,
+      "step": 792
+    }
+  ],
+  "max_steps": 1188,
+  "num_train_epochs": 3,
+  "total_flos": 2.100136125963174e+18,
+  "trial_name": null,
+  "trial_params": null
+}