diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4813 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 798,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.4844,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.4482,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 1.385,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.2781,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.1396,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.6303,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.4753,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.1886,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.1805,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.139,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.0784,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.0639,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.0573,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.0564,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0517,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0617,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 0.0612,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0608,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.0478,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.0492,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.0618,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.0552,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 0.0555,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999991762655447e-05,
+      "loss": 0.0496,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999670507574945e-05,
+      "loss": 0.0537,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999258647132645e-05,
+      "loss": 0.0474,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999868205201284e-05,
+      "loss": 0.0552,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997940731714748e-05,
+      "loss": 0.0499,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997034698451396e-05,
+      "loss": 0.0564,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99959639671494e-05,
+      "loss": 0.0437,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994728555448723e-05,
+      "loss": 0.0514,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993328483702393e-05,
+      "loss": 0.0449,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999176377497616e-05,
+      "loss": 0.0532,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990034455048098e-05,
+      "loss": 0.0491,
+      "step": 35
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998814055240823e-05,
+      "loss": 0.0465,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998608209825801e-05,
+      "loss": 0.0595,
+      "step": 37
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983859126509827e-05,
+      "loss": 0.0521,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998147167378645e-05,
+      "loss": 0.0569,
+      "step": 39
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978919779420427e-05,
+      "loss": 0.0535,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976203485453415e-05,
+      "loss": 0.0492,
+      "step": 41
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973322836635517e-05,
+      "loss": 0.0485,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997027788042453e-05,
+      "loss": 0.0528,
+      "step": 43
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996706866698515e-05,
+      "loss": 0.0517,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963695249188185e-05,
+      "loss": 0.0524,
+      "step": 45
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960157682609634e-05,
+      "loss": 0.0529,
+      "step": 46
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956456025529808e-05,
+      "loss": 0.0565,
+      "step": 47
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995259033893236e-05,
+      "loss": 0.0609,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948560686503275e-05,
+      "loss": 0.0578,
+      "step": 49
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994436713462982e-05,
+      "loss": 0.0504,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940009752399462e-05,
+      "loss": 0.0519,
+      "step": 51
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935488611598716e-05,
+      "loss": 0.053,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993080378671197e-05,
+      "loss": 0.0515,
+      "step": 53
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9925955354920265e-05,
+      "loss": 0.0538,
+      "step": 54
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99209433961e-05,
+      "loss": 0.0527,
+      "step": 55
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915767992821642e-05,
+      "loss": 0.0507,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910429230348348e-05,
+      "loss": 0.0543,
+      "step": 57
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904927196634572e-05,
+      "loss": 0.0487,
+      "step": 58
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9899261982324608e-05,
+      "loss": 0.0515,
+      "step": 59
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9893433680751105e-05,
+      "loss": 0.0503,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988744238793351e-05,
+      "loss": 0.0474,
+      "step": 61
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9881288202576518e-05,
+      "loss": 0.0438,
+      "step": 62
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9874971226068417e-05,
+      "loss": 0.052,
+      "step": 63
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986849156247943e-05,
+      "loss": 0.0526,
+      "step": 64
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986184931856e-05,
+      "loss": 0.0541,
+      "step": 65
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985504460373903e-05,
+      "loss": 0.0512,
+      "step": 66
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.0499,
+      "step": 67
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9840948212489527e-05,
+      "loss": 0.0497,
+      "step": 68
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983365676829466e-05,
+      "loss": 0.0518,
+      "step": 69
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826203317661758e-05,
+      "loss": 0.0484,
+      "step": 70
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9818587983384098e-05,
+      "loss": 0.0509,
+      "step": 71
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9810810890921943e-05,
+      "loss": 0.0555,
+      "step": 72
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980287216840048e-05,
+      "loss": 0.0573,
+      "step": 73
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979477194660769e-05,
+      "loss": 0.0536,
+      "step": 74
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9786510358992213e-05,
+      "loss": 0.0506,
+      "step": 75
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977808754166113e-05,
+      "loss": 0.0559,
+      "step": 76
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9769503633377745e-05,
+      "loss": 0.0495,
+      "step": 77
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9760758775559275e-05,
+      "loss": 0.0583,
+      "step": 78
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975185311227453e-05,
+      "loss": 0.0562,
+      "step": 79
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9742786790241548e-05,
+      "loss": 0.0513,
+      "step": 80
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9733559958825167e-05,
+      "loss": 0.0506,
+      "step": 81
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9724172770034566e-05,
+      "loss": 0.0515,
+      "step": 82
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971462537852076e-05,
+      "loss": 0.054,
+      "step": 83
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9704917941574053e-05,
+      "loss": 0.0498,
+      "step": 84
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9695050619121457e-05,
+      "loss": 0.0508,
+      "step": 85
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9685023573724036e-05,
+      "loss": 0.0569,
+      "step": 86
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9674836970574253e-05,
+      "loss": 0.0477,
+      "step": 87
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9664490977493223e-05,
+      "loss": 0.0495,
+      "step": 88
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9653985764927964e-05,
+      "loss": 0.0496,
+      "step": 89
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643321505948588e-05,
+      "loss": 0.05,
+      "step": 90
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963249837624545e-05,
+      "loss": 0.0583,
+      "step": 91
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962151655412624e-05,
+      "loss": 0.0581,
+      "step": 92
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9610376220513067e-05,
+      "loss": 0.0554,
+      "step": 93
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9599077558939468e-05,
+      "loss": 0.0516,
+      "step": 94
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9587620755547375e-05,
+      "loss": 0.0583,
+      "step": 95
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957600599908406e-05,
+      "loss": 0.0529,
+      "step": 96
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956423348089903e-05,
+      "loss": 0.0495,
+      "step": 97
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9552303394940862e-05,
+      "loss": 0.0456,
+      "step": 98
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954021593775401e-05,
+      "loss": 0.0533,
+      "step": 99
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952797130847557e-05,
+      "loss": 0.0536,
+      "step": 100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9515569708832012e-05,
+      "loss": 0.0504,
+      "step": 101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9503011343135828e-05,
+      "loss": 0.0556,
+      "step": 102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9490296418282187e-05,
+      "loss": 0.0552,
+      "step": 103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947742514374553e-05,
+      "loss": 0.0507,
+      "step": 104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9464397731576093e-05,
+      "loss": 0.0567,
+      "step": 105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9451214396396453e-05,
+      "loss": 0.0501,
+      "step": 106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9437875355397953e-05,
+      "loss": 0.0479,
+      "step": 107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9424380828337146e-05,
+      "loss": 0.0493,
+      "step": 108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941073103753217e-05,
+      "loss": 0.0442,
+      "step": 109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.0627,
+      "step": 110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938296656674817e-05,
+      "loss": 0.0529,
+      "step": 111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9368852344180168e-05,
+      "loss": 0.0522,
+      "step": 112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9354583772682512e-05,
+      "loss": 0.0474,
+      "step": 113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9340161087325483e-05,
+      "loss": 0.0456,
+      "step": 114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9325584525718334e-05,
+      "loss": 0.0523,
+      "step": 115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9310854328005383e-05,
+      "loss": 0.0501,
+      "step": 116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9295970736862063e-05,
+      "loss": 0.0543,
+      "step": 117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9280933997490912e-05,
+      "loss": 0.0515,
+      "step": 118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9265744357617532e-05,
+      "loss": 0.0507,
+      "step": 119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9250402067486523e-05,
+      "loss": 0.0494,
+      "step": 120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9234907379857336e-05,
+      "loss": 0.049,
+      "step": 121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9219260550000144e-05,
+      "loss": 0.0547,
+      "step": 122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9203461835691596e-05,
+      "loss": 0.0441,
+      "step": 123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.91875114972106e-05,
+      "loss": 0.0502,
+      "step": 124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917140979733403e-05,
+      "loss": 0.0496,
+      "step": 125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9155157001332374e-05,
+      "loss": 0.0501,
+      "step": 126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.91387533769654e-05,
+      "loss": 0.0554,
+      "step": 127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9122199194477723e-05,
+      "loss": 0.0515,
+      "step": 128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9105494726594344e-05,
+      "loss": 0.0509,
+      "step": 129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9088640248516185e-05,
+      "loss": 0.0605,
+      "step": 130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9071636037915537e-05,
+      "loss": 0.0606,
+      "step": 131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905448237493147e-05,
+      "loss": 0.0541,
+      "step": 132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9037179542165255e-05,
+      "loss": 0.0489,
+      "step": 133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9019727824675686e-05,
+      "loss": 0.061,
+      "step": 134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9002127509974376e-05,
+      "loss": 0.0599,
+      "step": 135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8984378888021045e-05,
+      "loss": 0.0594,
+      "step": 136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8966482251218716e-05,
+      "loss": 0.0473,
+      "step": 137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894843789440892e-05,
+      "loss": 0.0465,
+      "step": 138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8930246114866825e-05,
+      "loss": 0.0536,
+      "step": 139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8911907212296343e-05,
+      "loss": 0.0448,
+      "step": 140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889342148882519e-05,
+      "loss": 0.0597,
+      "step": 141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8874789248999915e-05,
+      "loss": 0.0589,
+      "step": 142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8856010799780883e-05,
+      "loss": 0.0495,
+      "step": 143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8837086450537195e-05,
+      "loss": 0.0566,
+      "step": 144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8818016513041628e-05,
+      "loss": 0.0519,
+      "step": 145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879880130146547e-05,
+      "loss": 0.0563,
+      "step": 146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877944113237336e-05,
+      "loss": 0.0528,
+      "step": 147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8759936324718068e-05,
+      "loss": 0.0485,
+      "step": 148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8740287199835234e-05,
+      "loss": 0.0484,
+      "step": 149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872049408143808e-05,
+      "loss": 0.0421,
+      "step": 150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8700557295612074e-05,
+      "loss": 0.0434,
+      "step": 151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8680477170809573e-05,
+      "loss": 0.0509,
+      "step": 152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.0514,
+      "step": 153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8639888229886342e-05,
+      "loss": 0.0535,
+      "step": 154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8619380082455796e-05,
+      "loss": 0.056,
+      "step": 155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8598729933418102e-05,
+      "loss": 0.0541,
+      "step": 156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857793812297804e-05,
+      "loss": 0.0526,
+      "step": 157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855700499367423e-05,
+      "loss": 0.0518,
+      "step": 158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8535930890373467e-05,
+      "loss": 0.0498,
+      "step": 159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851471616026505e-05,
+      "loss": 0.0505,
+      "step": 160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.849336115285506e-05,
+      "loss": 0.0512,
+      "step": 161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8471866219960604e-05,
+      "loss": 0.0513,
+      "step": 162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.845023171570403e-05,
+      "loss": 0.0547,
+      "step": 163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8428457996507053e-05,
+      "loss": 0.053,
+      "step": 164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840654542108494e-05,
+      "loss": 0.0568,
+      "step": 165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8384494350440553e-05,
+      "loss": 0.0504,
+      "step": 166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.836230514785843e-05,
+      "loss": 0.0581,
+      "step": 167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833997817889878e-05,
+      "loss": 0.0457,
+      "step": 168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831751381139148e-05,
+      "loss": 0.0454,
+      "step": 169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8294912415429995e-05,
+      "loss": 0.0644,
+      "step": 170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.82721743633653e-05,
+      "loss": 0.0517,
+      "step": 171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8249300029799735e-05,
+      "loss": 0.0463,
+      "step": 172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8226289791580833e-05,
+      "loss": 0.0507,
+      "step": 173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820314402779511e-05,
+      "loss": 0.0514,
+      "step": 174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8179863119761837e-05,
+      "loss": 0.0537,
+      "step": 175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815644745102673e-05,
+      "loss": 0.0503,
+      "step": 176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8132897407355657e-05,
+      "loss": 0.0536,
+      "step": 177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.810921337672826e-05,
+      "loss": 0.0514,
+      "step": 178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.808539574933158e-05,
+      "loss": 0.0492,
+      "step": 179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.806144491755363e-05,
+      "loss": 0.0553,
+      "step": 180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803736127597691e-05,
+      "loss": 0.0472,
+      "step": 181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8013145221371937e-05,
+      "loss": 0.0595,
+      "step": 182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.798879715269067e-05,
+      "loss": 0.0537,
+      "step": 183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7964317471059982e-05,
+      "loss": 0.0522,
+      "step": 184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7939706579775015e-05,
+      "loss": 0.0579,
+      "step": 185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7914964884292543e-05,
+      "loss": 0.0515,
+      "step": 186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7890092792224316e-05,
+      "loss": 0.0451,
+      "step": 187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7865090713330313e-05,
+      "loss": 0.0423,
+      "step": 188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7839959059512016e-05,
+      "loss": 0.0521,
+      "step": 189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7814698244805605e-05,
+      "loss": 0.0572,
+      "step": 190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778930868537515e-05,
+      "loss": 0.0589,
+      "step": 191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7763790799505746e-05,
+      "loss": 0.0506,
+      "step": 192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.773814500759663e-05,
+      "loss": 0.0492,
+      "step": 193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.771237173215426e-05,
+      "loss": 0.0519,
+      "step": 194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7686471397785322e-05,
+      "loss": 0.0441,
+      "step": 195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.055,
+      "step": 196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763429126115382e-05,
+      "loss": 0.0509,
+      "step": 197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.760801231854278e-05,
+      "loss": 0.0502,
+      "step": 198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7581608036294077e-05,
+      "loss": 0.0466,
+      "step": 199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7555078849410044e-05,
+      "loss": 0.0492,
+      "step": 200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7528425194950794e-05,
+      "loss": 0.0527,
+      "step": 201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7501647512026996e-05,
+      "loss": 0.0546,
+      "step": 202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747474624179265e-05,
+      "loss": 0.053,
+      "step": 203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.744772182743782e-05,
+      "loss": 0.0558,
+      "step": 204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.742057471418133e-05,
+      "loss": 0.0454,
+      "step": 205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7393305349263433e-05,
+      "loss": 0.0601,
+      "step": 206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736591418193844e-05,
+      "loss": 0.0512,
+      "step": 207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733840166346731e-05,
+      "loss": 0.0605,
+      "step": 208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7310768247110234e-05,
+      "loss": 0.0501,
+      "step": 209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.728301438811916e-05,
+      "loss": 0.0533,
+      "step": 210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7255140543730286e-05,
+      "loss": 0.0487,
+      "step": 211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7227147173156525e-05,
+      "loss": 0.0595,
+      "step": 212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7199034737579962e-05,
+      "loss": 0.052,
+      "step": 213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7170803700144227e-05,
+      "loss": 0.0548,
+      "step": 214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.714245452594689e-05,
+      "loss": 0.0534,
+      "step": 215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.711398768203178e-05,
+      "loss": 0.0476,
+      "step": 216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.70854036373813e-05,
+      "loss": 0.0526,
+      "step": 217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7056702862908704e-05,
+      "loss": 0.0569,
+      "step": 218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7027885831450318e-05,
+      "loss": 0.0532,
+      "step": 219
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6998953017757787e-05,
+      "loss": 0.0517,
+      "step": 220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6969904898490215e-05,
+      "loss": 0.0464,
+      "step": 221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6940741952206342e-05,
+      "loss": 0.0602,
+      "step": 222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6911464659356633e-05,
+      "loss": 0.0507,
+      "step": 223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6882073502275394e-05,
+      "loss": 0.0513,
+      "step": 224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6852568965172794e-05,
+      "loss": 0.0517,
+      "step": 225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.682295153412691e-05,
+      "loss": 0.0476,
+      "step": 226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6793221697075716e-05,
+      "loss": 0.0498,
+      "step": 227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.676337994380903e-05,
+      "loss": 0.0566,
+      "step": 228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673342676596046e-05,
+      "loss": 0.048,
+      "step": 229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.67033626569993e-05,
+      "loss": 0.041,
+      "step": 230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6673188112222394e-05,
+      "loss": 0.0521,
+      "step": 231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6642903628745995e-05,
+      "loss": 0.0526,
+      "step": 232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6612509705497545e-05,
+      "loss": 0.0523,
+      "step": 233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.658200684320748e-05,
+      "loss": 0.0549,
+      "step": 234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6551395544400982e-05,
+      "loss": 0.0516,
+      "step": 235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.652067631338967e-05,
+      "loss": 0.0502,
+      "step": 236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6489849656263336e-05,
+      "loss": 0.0468,
+      "step": 237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6458916080881566e-05,
+      "loss": 0.0498,
+      "step": 238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.0502,
+      "step": 239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6396730215588913e-05,
+      "loss": 0.0488,
+      "step": 240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6365478950170834e-05,
+      "loss": 0.0506,
+      "step": 241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6334122815466035e-05,
+      "loss": 0.0515,
+      "step": 242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.630266232805709e-05,
+      "loss": 0.0473,
+      "step": 243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6271098006245742e-05,
+      "loss": 0.0568,
+      "step": 244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.623943037004439e-05,
+      "loss": 0.0524,
+      "step": 245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6207659941167485e-05,
+      "loss": 0.0529,
+      "step": 246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6175787243022975e-05,
+      "loss": 0.0485,
+      "step": 247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6143812800703646e-05,
+      "loss": 0.0522,
+      "step": 248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6111737140978495e-05,
+      "loss": 0.0505,
+      "step": 249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6079560792284046e-05,
+      "loss": 0.0505,
+      "step": 250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6047284284715642e-05,
+      "loss": 0.0555,
+      "step": 251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6014908150018703e-05,
+      "loss": 0.0484,
+      "step": 252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5982432921579994e-05,
+      "loss": 0.0578,
+      "step": 253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5949859134418798e-05,
+      "loss": 0.0506,
+      "step": 254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591718732517814e-05,
+      "loss": 0.0558,
+      "step": 255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5884418032115907e-05,
+      "loss": 0.0497,
+      "step": 256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5851551795096026e-05,
+      "loss": 0.0605,
+      "step": 257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.581858915557953e-05,
+      "loss": 0.0565,
+      "step": 258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5785530656615657e-05,
+      "loss": 0.0486,
+      "step": 259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.57523768428329e-05,
+      "loss": 0.0504,
+      "step": 260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.571912826043003e-05,
+      "loss": 0.0514,
+      "step": 261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5685785457167114e-05,
+      "loss": 0.047,
+      "step": 262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5652348982356463e-05,
+      "loss": 0.0442,
+      "step": 263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5618819386853607e-05,
+      "loss": 0.0545,
+      "step": 264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5585197223048206e-05,
+      "loss": 0.0437,
+      "step": 265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5551483044854954e-05,
+      "loss": 0.0426,
+      "step": 266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.551767740770446e-05,
+      "loss": 0.0493,
+      "step": 267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5483780868534087e-05,
+      "loss": 0.059,
+      "step": 268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.544979398577877e-05,
+      "loss": 0.0479,
+      "step": 269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.541571731936185e-05,
+      "loss": 0.0589,
+      "step": 270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5381551430685796e-05,
+      "loss": 0.048,
+      "step": 271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534729688262302e-05,
+      "loss": 0.0573,
+      "step": 272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5312954239506536e-05,
+      "loss": 0.0591,
+      "step": 273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527852406712072e-05,
+      "loss": 0.0486,
+      "step": 274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5244006932691954e-05,
+      "loss": 0.0525,
+      "step": 275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5209403404879305e-05,
+      "loss": 0.0583,
+      "step": 276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5174714053765125e-05,
+      "loss": 0.0545,
+      "step": 277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5139939450845699e-05,
+      "loss": 0.0578,
+      "step": 278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5105080169021792e-05,
+      "loss": 0.0503,
+      "step": 279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5070136782589236e-05,
+      "loss": 0.0475,
+      "step": 280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5035109867229458e-05,
+      "loss": 0.0512,
+      "step": 281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.06,
+      "step": 282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.496480775932501e-05,
+      "loss": 0.0608,
+      "step": 283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4929533724985712e-05,
+      "loss": 0.0512,
+      "step": 284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4894178478110856e-05,
+      "loss": 0.0507,
+      "step": 285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.485874260116714e-05,
+      "loss": 0.0485,
+      "step": 286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4823226677949623e-05,
+      "loss": 0.0574,
+      "step": 287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4787631293572094e-05,
+      "loss": 0.0555,
+      "step": 288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4751957034457447e-05,
+      "loss": 0.0537,
+      "step": 289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4716204488328007e-05,
+      "loss": 0.0607,
+      "step": 290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4680374244195861e-05,
+      "loss": 0.0508,
+      "step": 291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4644466892353142e-05,
+      "loss": 0.0589,
+      "step": 292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4608483024362303e-05,
+      "loss": 0.0541,
+      "step": 293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4572423233046386e-05,
+      "loss": 0.051,
+      "step": 294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4536288112479242e-05,
+      "loss": 0.0472,
+      "step": 295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4500078257975748e-05,
+      "loss": 0.047,
+      "step": 296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4463794266081994e-05,
+      "loss": 0.0585,
+      "step": 297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4427436734565475e-05,
+      "loss": 0.052,
+      "step": 298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4391006262405215e-05,
+      "loss": 0.0446,
+      "step": 299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4354503449781914e-05,
+      "loss": 0.0488,
+      "step": 300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4317928898068067e-05,
+      "loss": 0.0478,
+      "step": 301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.428128320981804e-05,
+      "loss": 0.0514,
+      "step": 302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4244566988758152e-05,
+      "loss": 0.0439,
+      "step": 303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4207780839776735e-05,
+      "loss": 0.049,
+      "step": 304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4170925368914152e-05,
+      "loss": 0.0577,
+      "step": 305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4134001183352833e-05,
+      "loss": 0.0538,
+      "step": 306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4097008891407246e-05,
+      "loss": 0.0454,
+      "step": 307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4059949102513914e-05,
+      "loss": 0.0475,
+      "step": 308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4022822427221325e-05,
+      "loss": 0.0493,
+      "step": 309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3985629477179915e-05,
+      "loss": 0.0554,
+      "step": 310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.394837086513198e-05,
+      "loss": 0.053,
+      "step": 311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.391104720490156e-05,
+      "loss": 0.0564,
+      "step": 312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3873659111384363e-05,
+      "loss": 0.0562,
+      "step": 313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3836207200537597e-05,
+      "loss": 0.0608,
+      "step": 314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3798692089369855e-05,
+      "loss": 0.0474,
+      "step": 315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.376111439593093e-05,
+      "loss": 0.0425,
+      "step": 316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3723474739301636e-05,
+      "loss": 0.0504,
+      "step": 317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.368577373958362e-05,
+      "loss": 0.0552,
+      "step": 318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3648012017889122e-05,
+      "loss": 0.0571,
+      "step": 319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3610190196330777e-05,
+      "loss": 0.0513,
+      "step": 320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3572308898011328e-05,
+      "loss": 0.0543,
+      "step": 321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3534368747013396e-05,
+      "loss": 0.0489,
+      "step": 322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3496370368389167e-05,
+      "loss": 0.0529,
+      "step": 323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3458314388150115e-05,
+      "loss": 0.0531,
+      "step": 324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.0536,
+      "step": 325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3382032131607967e-05,
+      "loss": 0.0576,
+      "step": 326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3343807112031329e-05,
+      "loss": 0.0526,
+      "step": 327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3305527004272087e-05,
+      "loss": 0.0602,
+      "step": 328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3267192438983118e-05,
+      "loss": 0.0538,
+      "step": 329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3228804047714462e-05,
+      "loss": 0.0488,
+      "step": 330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3190362462902937e-05,
+      "loss": 0.0509,
+      "step": 331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.31518683178617e-05,
+      "loss": 0.059,
+      "step": 332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3113322246769817e-05,
+      "loss": 0.052,
+      "step": 333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3074724884661833e-05,
+      "loss": 0.0458,
+      "step": 334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3036076867417286e-05,
+      "loss": 0.0485,
+      "step": 335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2997378831750242e-05,
+      "loss": 0.0499,
+      "step": 336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958631415198815e-05,
+      "loss": 0.0652,
+      "step": 337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2919835256114639e-05,
+      "loss": 0.0521,
+      "step": 338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2880990993652379e-05,
+      "loss": 0.0497,
+      "step": 339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2842099267759176e-05,
+      "loss": 0.0537,
+      "step": 340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2803160719164127e-05,
+      "loss": 0.0461,
+      "step": 341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2764175989367717e-05,
+      "loss": 0.0495,
+      "step": 342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2725145720631243e-05,
+      "loss": 0.0582,
+      "step": 343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2686070555966255e-05,
+      "loss": 0.0594,
+      "step": 344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2646951139123935e-05,
+      "loss": 0.0497,
+      "step": 345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2607788114584522e-05,
+      "loss": 0.0629,
+      "step": 346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2568582127546663e-05,
+      "loss": 0.0481,
+      "step": 347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2529333823916807e-05,
+      "loss": 0.0585,
+      "step": 348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2490043850298559e-05,
+      "loss": 0.0505,
+      "step": 349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2450712853982014e-05,
+      "loss": 0.0583,
+      "step": 350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.241134148293311e-05,
+      "loss": 0.0512,
+      "step": 351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2371930385782944e-05,
+      "loss": 0.0487,
+      "step": 352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2332480211817093e-05,
+      "loss": 0.0526,
+      "step": 353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2292991610964902e-05,
+      "loss": 0.0482,
+      "step": 354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2253465233788794e-05,
+      "loss": 0.0563,
+      "step": 355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2213901731473555e-05,
+      "loss": 0.0613,
+      "step": 356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2174301755815572e-05,
+      "loss": 0.05,
+      "step": 357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2134665959212138e-05,
+      "loss": 0.0519,
+      "step": 358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2094994994650682e-05,
+      "loss": 0.0653,
+      "step": 359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2055289515698008e-05,
+      "loss": 0.0474,
+      "step": 360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2015550176489539e-05,
+      "loss": 0.0542,
+      "step": 361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1975777631718533e-05,
+      "loss": 0.0536,
+      "step": 362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1935972536625302e-05,
+      "loss": 0.0534,
+      "step": 363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.189613554698641e-05,
+      "loss": 0.0461,
+      "step": 364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1856267319103878e-05,
+      "loss": 0.0531,
+      "step": 365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1816368509794365e-05,
+      "loss": 0.0441,
+      "step": 366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1776439776378353e-05,
+      "loss": 0.0497,
+      "step": 367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.0597,
+      "step": 368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1696495168962848e-05,
+      "loss": 0.0518,
+      "step": 369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1656480612025914e-05,
+      "loss": 0.0516,
+      "step": 370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1616438765085883e-05,
+      "loss": 0.0556,
+      "step": 371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1576370287819737e-05,
+      "loss": 0.0578,
+      "step": 372
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1536275840343186e-05,
+      "loss": 0.0514,
+      "step": 373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.149615608319978e-05,
+      "loss": 0.0537,
+      "step": 374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1456011677350052e-05,
+      "loss": 0.0512,
+      "step": 375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.14158432841606e-05,
+      "loss": 0.0572,
+      "step": 376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.137565156539322e-05,
+      "loss": 0.0502,
+      "step": 377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.133543718319398e-05,
+      "loss": 0.051,
+      "step": 378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1295200800082326e-05,
+      "loss": 0.0539,
+      "step": 379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1254943078940161e-05,
+      "loss": 0.0546,
+      "step": 380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1214664683000927e-05,
+      "loss": 0.05,
+      "step": 381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1174366275838664e-05,
+      "loss": 0.0525,
+      "step": 382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1134048521357117e-05,
+      "loss": 0.0485,
+      "step": 383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1093712083778748e-05,
+      "loss": 0.0516,
+      "step": 384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1053357627633823e-05,
+      "loss": 0.0546,
+      "step": 385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1012985817749465e-05,
+      "loss": 0.0526,
+      "step": 386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0972597319238692e-05,
+      "loss": 0.0548,
+      "step": 387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.093219279748946e-05,
+      "loss": 0.0586,
+      "step": 388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0891772918153696e-05,
+      "loss": 0.0557,
+      "step": 389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0851338347136358e-05,
+      "loss": 0.0474,
+      "step": 390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0810889750584426e-05,
+      "loss": 0.0461,
+      "step": 391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.077042779487595e-05,
+      "loss": 0.0479,
+      "step": 392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0729953146609076e-05,
+      "loss": 0.0544,
+      "step": 393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.068946647259105e-05,
+      "loss": 0.0491,
+      "step": 394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0648968439827242e-05,
+      "loss": 0.0577,
+      "step": 395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.060845971551014e-05,
+      "loss": 0.0496,
+      "step": 396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0567940967008397e-05,
+      "loss": 0.0455,
+      "step": 397
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.052741286185579e-05,
+      "loss": 0.0474,
+      "step": 398
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0486876067740253e-05,
+      "loss": 0.0541,
+      "step": 399
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0446331252492864e-05,
+      "loss": 0.053,
+      "step": 400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0405779084076857e-05,
+      "loss": 0.0504,
+      "step": 401
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0365220230576592e-05,
+      "loss": 0.0569,
+      "step": 402
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0324655360186568e-05,
+      "loss": 0.0439,
+      "step": 403
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0284085141200424e-05,
+      "loss": 0.0484,
+      "step": 404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0243510241999898e-05,
+      "loss": 0.0516,
+      "step": 405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0202931331043839e-05,
+      "loss": 0.0474,
+      "step": 406
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0162349076857191e-05,
+      "loss": 0.0484,
+      "step": 407
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0121764148019977e-05,
+      "loss": 0.0556,
+      "step": 408
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.008117721315628e-05,
+      "loss": 0.0567,
+      "step": 409
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.004058894092323e-05,
+      "loss": 0.0528,
+      "step": 410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.0462,
+      "step": 411
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.95941105907677e-06,
+      "loss": 0.054,
+      "step": 412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.918822786843725e-06,
+      "loss": 0.05,
+      "step": 413
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.878235851980027e-06,
+      "loss": 0.0512,
+      "step": 414
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.83765092314281e-06,
+      "loss": 0.0548,
+      "step": 415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.797068668956163e-06,
+      "loss": 0.0499,
+      "step": 416
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.756489758000105e-06,
+      "loss": 0.0484,
+      "step": 417
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.715914858799576e-06,
+      "loss": 0.0466,
+      "step": 418
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.675344639813434e-06,
+      "loss": 0.0591,
+      "step": 419
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.634779769423412e-06,
+      "loss": 0.0478,
+      "step": 420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.59422091592315e-06,
+      "loss": 0.0508,
+      "step": 421
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.553668747507139e-06,
+      "loss": 0.0492,
+      "step": 422
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.51312393225975e-06,
+      "loss": 0.0499,
+      "step": 423
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.472587138144215e-06,
+      "loss": 0.0469,
+      "step": 424
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.432059032991607e-06,
+      "loss": 0.0447,
+      "step": 425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.391540284489862e-06,
+      "loss": 0.0557,
+      "step": 426
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.351031560172765e-06,
+      "loss": 0.0467,
+      "step": 427
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.310533527408953e-06,
+      "loss": 0.0467,
+      "step": 428
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.270046853390924e-06,
+      "loss": 0.0438,
+      "step": 429
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.229572205124052e-06,
+      "loss": 0.0545,
+      "step": 430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.189110249415577e-06,
+      "loss": 0.0496,
+      "step": 431
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.148661652863644e-06,
+      "loss": 0.0532,
+      "step": 432
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.108227081846306e-06,
+      "loss": 0.0571,
+      "step": 433
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.067807202510543e-06,
+      "loss": 0.0527,
+      "step": 434
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.02740268076131e-06,
+      "loss": 0.0491,
+      "step": 435
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.987014182250538e-06,
+      "loss": 0.0544,
+      "step": 436
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.946642372366182e-06,
+      "loss": 0.0541,
+      "step": 437
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.906287916221259e-06,
+      "loss": 0.0474,
+      "step": 438
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.865951478642888e-06,
+      "loss": 0.0507,
+      "step": 439
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.825633724161336e-06,
+      "loss": 0.0476,
+      "step": 440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.785335316999078e-06,
+      "loss": 0.0469,
+      "step": 441
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.74505692105984e-06,
+      "loss": 0.0576,
+      "step": 442
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.704799199917674e-06,
+      "loss": 0.0568,
+      "step": 443
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.664562816806022e-06,
+      "loss": 0.0535,
+      "step": 444
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.624348434606782e-06,
+      "loss": 0.0503,
+      "step": 445
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.584156715839402e-06,
+      "loss": 0.0611,
+      "step": 446
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.543988322649954e-06,
+      "loss": 0.0587,
+      "step": 447
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.503843916800221e-06,
+      "loss": 0.0549,
+      "step": 448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.463724159656816e-06,
+      "loss": 0.0437,
+      "step": 449
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.423629712180265e-06,
+      "loss": 0.0517,
+      "step": 450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.38356123491412e-06,
+      "loss": 0.0661,
+      "step": 451
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.343519387974091e-06,
+      "loss": 0.0547,
+      "step": 452
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.303504831037154e-06,
+      "loss": 0.0439,
+      "step": 453
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.048,
+      "step": 454
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.223560223621652e-06,
+      "loss": 0.0563,
+      "step": 455
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.183631490205636e-06,
+      "loss": 0.0499,
+      "step": 456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.143732680896124e-06,
+      "loss": 0.0517,
+      "step": 457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.103864453013593e-06,
+      "loss": 0.0565,
+      "step": 458
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.064027463374702e-06,
+      "loss": 0.0532,
+      "step": 459
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.024222368281469e-06,
+      "loss": 0.0574,
+      "step": 460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.984449823510468e-06,
+      "loss": 0.0457,
+      "step": 461
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.944710484301995e-06,
+      "loss": 0.0508,
+      "step": 462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.90500500534932e-06,
+      "loss": 0.0487,
+      "step": 463
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.865334040787866e-06,
+      "loss": 0.0457,
+      "step": 464
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.825698244184432e-06,
+      "loss": 0.0508,
+      "step": 465
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.786098268526447e-06,
+      "loss": 0.0505,
+      "step": 466
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.746534766211207e-06,
+      "loss": 0.0569,
+      "step": 467
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.707008389035102e-06,
+      "loss": 0.0517,
+      "step": 468
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.667519788182912e-06,
+      "loss": 0.0558,
+      "step": 469
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.628069614217058e-06,
+      "loss": 0.0516,
+      "step": 470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.588658517066893e-06,
+      "loss": 0.0569,
+      "step": 471
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.54928714601799e-06,
+      "loss": 0.0514,
+      "step": 472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.509956149701444e-06,
+      "loss": 0.0546,
+      "step": 473
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.470666176083193e-06,
+      "loss": 0.0453,
+      "step": 474
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.431417872453339e-06,
+      "loss": 0.0517,
+      "step": 475
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3922118854154815e-06,
+      "loss": 0.0445,
+      "step": 476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3530488608760645e-06,
+      "loss": 0.0526,
+      "step": 477
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.31392944403375e-06,
+      "loss": 0.0512,
+      "step": 478
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2748542793687595e-06,
+      "loss": 0.0471,
+      "step": 479
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.235824010632284e-06,
+      "loss": 0.0573,
+      "step": 480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.196839280835876e-06,
+      "loss": 0.0438,
+      "step": 481
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.157900732240828e-06,
+      "loss": 0.0536,
+      "step": 482
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.119009006347625e-06,
+      "loss": 0.0477,
+      "step": 483
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.080164743885364e-06,
+      "loss": 0.0493,
+      "step": 484
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.041368584801187e-06,
+      "loss": 0.0495,
+      "step": 485
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.002621168249759e-06,
+      "loss": 0.0574,
+      "step": 486
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.963923132582716e-06,
+      "loss": 0.0479,
+      "step": 487
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.925275115338168e-06,
+      "loss": 0.0532,
+      "step": 488
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.886677753230184e-06,
+      "loss": 0.0509,
+      "step": 489
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.848131682138304e-06,
+      "loss": 0.0483,
+      "step": 490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8096375370970625e-06,
+      "loss": 0.05,
+      "step": 491
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.771195952285541e-06,
+      "loss": 0.0502,
+      "step": 492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7328075610168855e-06,
+      "loss": 0.053,
+      "step": 493
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.694472995727914e-06,
+      "loss": 0.051,
+      "step": 494
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.656192887968675e-06,
+      "loss": 0.0482,
+      "step": 495
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.617967868392036e-06,
+      "loss": 0.0485,
+      "step": 496
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.0553,
+      "step": 497
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5416856118498874e-06,
+      "loss": 0.0526,
+      "step": 498
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.503629631610837e-06,
+      "loss": 0.0472,
+      "step": 499
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4656312529866086e-06,
+      "loss": 0.0446,
+      "step": 500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.427691101988673e-06,
+      "loss": 0.0525,
+      "step": 501
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.389809803669226e-06,
+      "loss": 0.0542,
+      "step": 502
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.35198798211088e-06,
+      "loss": 0.0467,
+      "step": 503
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.314226260416383e-06,
+      "loss": 0.0523,
+      "step": 504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.276525260698364e-06,
+      "loss": 0.0539,
+      "step": 505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.238885604069075e-06,
+      "loss": 0.0506,
+      "step": 506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2013079106301454e-06,
+      "loss": 0.049,
+      "step": 507
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.163792799462404e-06,
+      "loss": 0.0495,
+      "step": 508
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.126340888615642e-06,
+      "loss": 0.0601,
+      "step": 509
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.088952795098442e-06,
+      "loss": 0.0543,
+      "step": 510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.05162913486802e-06,
+      "loss": 0.054,
+      "step": 511
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.014370522820085e-06,
+      "loss": 0.0464,
+      "step": 512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.977177572778679e-06,
+      "loss": 0.0487,
+      "step": 513
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9400508974860895e-06,
+      "loss": 0.0463,
+      "step": 514
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.902991108592755e-06,
+      "loss": 0.05,
+      "step": 515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8659988166471715e-06,
+      "loss": 0.0534,
+      "step": 516
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8290746310858525e-06,
+      "loss": 0.0447,
+      "step": 517
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.792219160223268e-06,
+      "loss": 0.0624,
+      "step": 518
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.755433011241851e-06,
+      "loss": 0.0546,
+      "step": 519
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7187167901819665e-06,
+      "loss": 0.0542,
+      "step": 520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6820711019319364e-06,
+      "loss": 0.0516,
+      "step": 521
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.645496550218089e-06,
+      "loss": 0.0573,
+      "step": 522
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.608993737594791e-06,
+      "loss": 0.0532,
+      "step": 523
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.572563265434528e-06,
+      "loss": 0.0538,
+      "step": 524
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5362057339180075e-06,
+      "loss": 0.0494,
+      "step": 525
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.499921742024258e-06,
+      "loss": 0.0481,
+      "step": 526
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.46371188752076e-06,
+      "loss": 0.0557,
+      "step": 527
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.427576766953615e-06,
+      "loss": 0.0591,
+      "step": 528
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.391516975637699e-06,
+      "loss": 0.0589,
+      "step": 529
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3555331076468585e-06,
+      "loss": 0.0621,
+      "step": 530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3196257558041386e-06,
+      "loss": 0.0593,
+      "step": 531
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.283795511671994e-06,
+      "loss": 0.0517,
+      "step": 532
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.248042965542559e-06,
+      "loss": 0.0512,
+      "step": 533
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.212368706427913e-06,
+      "loss": 0.0496,
+      "step": 534
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.176773322050381e-06,
+      "loss": 0.0503,
+      "step": 535
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.141257398832863e-06,
+      "loss": 0.0545,
+      "step": 536
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.105821521889147e-06,
+      "loss": 0.0519,
+      "step": 537
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0704662750142875e-06,
+      "loss": 0.0567,
+      "step": 538
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.035192240674991e-06,
+      "loss": 0.0655,
+      "step": 539
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.0484,
+      "step": 540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.964890132770543e-06,
+      "loss": 0.0504,
+      "step": 541
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.929863217410768e-06,
+      "loss": 0.0582,
+      "step": 542
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.894919830978212e-06,
+      "loss": 0.0596,
+      "step": 543
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.860060549154302e-06,
+      "loss": 0.0621,
+      "step": 544
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8252859462348744e-06,
+      "loss": 0.0479,
+      "step": 545
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.790596595120699e-06,
+      "loss": 0.0544,
+      "step": 546
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7559930673080475e-06,
+      "loss": 0.06,
+      "step": 547
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.721475932879283e-06,
+      "loss": 0.0597,
+      "step": 548
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.687045760493468e-06,
+      "loss": 0.0561,
+      "step": 549
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.652703117376987e-06,
+      "loss": 0.0472,
+      "step": 550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.618448569314207e-06,
+      "loss": 0.0539,
+      "step": 551
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.584282680638155e-06,
+      "loss": 0.0485,
+      "step": 552
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.550206014221232e-06,
+      "loss": 0.0577,
+      "step": 553
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.51621913146592e-06,
+      "loss": 0.0545,
+      "step": 554
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.482322592295541e-06,
+      "loss": 0.0545,
+      "step": 555
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.448516955145048e-06,
+      "loss": 0.0486,
+      "step": 556
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.414802776951799e-06,
+      "loss": 0.0458,
+      "step": 557
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.381180613146396e-06,
+      "loss": 0.0542,
+      "step": 558
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.34765101764354e-06,
+      "loss": 0.0474,
+      "step": 559
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.314214542832889e-06,
+      "loss": 0.0452,
+      "step": 560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.280871739569972e-06,
+      "loss": 0.0483,
+      "step": 561
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2476231571671025e-06,
+      "loss": 0.0465,
+      "step": 562
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214469343384347e-06,
+      "loss": 0.0565,
+      "step": 563
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.181410844420473e-06,
+      "loss": 0.0502,
+      "step": 564
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.148448204903977e-06,
+      "loss": 0.0432,
+      "step": 565
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1155819678840935e-06,
+      "loss": 0.0486,
+      "step": 566
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.082812674821865e-06,
+      "loss": 0.0537,
+      "step": 567
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.050140865581205e-06,
+      "loss": 0.0552,
+      "step": 568
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.017567078420007e-06,
+      "loss": 0.0521,
+      "step": 569
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9850918499812976e-06,
+      "loss": 0.0475,
+      "step": 570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.952715715284363e-06,
+      "loss": 0.05,
+      "step": 571
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.920439207715955e-06,
+      "loss": 0.0483,
+      "step": 572
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.888262859021508e-06,
+      "loss": 0.0552,
+      "step": 573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8561871992963585e-06,
+      "loss": 0.058,
+      "step": 574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.824212756977027e-06,
+      "loss": 0.0614,
+      "step": 575
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7923400588325156e-06,
+      "loss": 0.0499,
+      "step": 576
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.760569629955614e-06,
+      "loss": 0.0425,
+      "step": 577
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7289019937542603e-06,
+      "loss": 0.0486,
+      "step": 578
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6973376719429134e-06,
+      "loss": 0.0493,
+      "step": 579
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.665877184533968e-06,
+      "loss": 0.0482,
+      "step": 580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6345210498291696e-06,
+      "loss": 0.0528,
+      "step": 581
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6032697844110896e-06,
+      "loss": 0.0511,
+      "step": 582
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.0518,
+      "step": 583
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5410839191184386e-06,
+      "loss": 0.0484,
+      "step": 584
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5101503437366678e-06,
+      "loss": 0.054,
+      "step": 585
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4793236866103296e-06,
+      "loss": 0.0508,
+      "step": 586
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4486044555990218e-06,
+      "loss": 0.0503,
+      "step": 587
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4179931567925216e-06,
+      "loss": 0.0505,
+      "step": 588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.387490294502457e-06,
+      "loss": 0.0514,
+      "step": 589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3570963712540083e-06,
+      "loss": 0.0492,
+      "step": 590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.326811887777607e-06,
+      "loss": 0.0451,
+      "step": 591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2966373430007047e-06,
+      "loss": 0.0586,
+      "step": 592
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.266573234039542e-06,
+      "loss": 0.0496,
+      "step": 593
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.236620056190972e-06,
+      "loss": 0.056,
+      "step": 594
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2067783029242872e-06,
+      "loss": 0.0495,
+      "step": 595
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1770484658730896e-06,
+      "loss": 0.0476,
+      "step": 596
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1474310348272084e-06,
+      "loss": 0.06,
+      "step": 597
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1179264977246106e-06,
+      "loss": 0.0554,
+      "step": 598
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0885353406433703e-06,
+      "loss": 0.0518,
+      "step": 599
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0592580477936606e-06,
+      "loss": 0.0531,
+      "step": 600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0300951015097867e-06,
+      "loss": 0.0521,
+      "step": 601
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.001046982242216e-06,
+      "loss": 0.0492,
+      "step": 602
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9721141685496825e-06,
+      "loss": 0.049,
+      "step": 603
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9432971370913e-06,
+      "loss": 0.0613,
+      "step": 604
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9145963626187014e-06,
+      "loss": 0.0507,
+      "step": 605
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8860123179682244e-06,
+      "loss": 0.0552,
+      "step": 606
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8575454740531117e-06,
+      "loss": 0.0466,
+      "step": 607
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8291962998557754e-06,
+      "loss": 0.0488,
+      "step": 608
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8009652624200436e-06,
+      "loss": 0.0498,
+      "step": 609
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.772852826843476e-06,
+      "loss": 0.0546,
+      "step": 610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7448594562697182e-06,
+      "loss": 0.0497,
+      "step": 611
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7169856118808414e-06,
+      "loss": 0.0475,
+      "step": 612
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6892317528897683e-06,
+      "loss": 0.0545,
+      "step": 613
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6615983365326926e-06,
+      "loss": 0.0513,
+      "step": 614
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.634085818061565e-06,
+      "loss": 0.056,
+      "step": 615
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.60669465073657e-06,
+      "loss": 0.0501,
+      "step": 616
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.579425285818671e-06,
+      "loss": 0.0498,
+      "step": 617
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5522781725621814e-06,
+      "loss": 0.0484,
+      "step": 618
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.525253758207353e-06,
+      "loss": 0.0516,
+      "step": 619
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.498352487973007e-06,
+      "loss": 0.0484,
+      "step": 620
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.471574805049206e-06,
+      "loss": 0.0551,
+      "step": 621
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.444921150589957e-06,
+      "loss": 0.0564,
+      "step": 622
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.418391963705927e-06,
+      "loss": 0.0531,
+      "step": 623
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3919876814572197e-06,
+      "loss": 0.051,
+      "step": 624
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.365708738846182e-06,
+      "loss": 0.0513,
+      "step": 625
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.0499,
+      "step": 626
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3135286022146785e-06,
+      "loss": 0.0529,
+      "step": 627
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.287628267845744e-06,
+      "loss": 0.0523,
+      "step": 628
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2618549924033707e-06,
+      "loss": 0.0605,
+      "step": 629
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2362092004942583e-06,
+      "loss": 0.056,
+      "step": 630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2106913146248544e-06,
+      "loss": 0.0565,
+      "step": 631
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.185301755194399e-06,
+      "loss": 0.0488,
+      "step": 632
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1600409404879875e-06,
+      "loss": 0.0508,
+      "step": 633
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1349092866696906e-06,
+      "loss": 0.0507,
+      "step": 634
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1099072077756864e-06,
+      "loss": 0.0506,
+      "step": 635
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.08503511570746e-06,
+      "loss": 0.0601,
+      "step": 636
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0602934202249913e-06,
+      "loss": 0.0535,
+      "step": 637
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0356825289400185e-06,
+      "loss": 0.0523,
+      "step": 638
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0112028473093294e-06,
+      "loss": 0.0434,
+      "step": 639
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986854778628067e-06,
+      "loss": 0.053,
+      "step": 640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.962638724023089e-06,
+      "loss": 0.0505,
+      "step": 641
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9385550824463727e-06,
+      "loss": 0.0534,
+      "step": 642
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9146042506684227e-06,
+      "loss": 0.0505,
+      "step": 643
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.890786623271743e-06,
+      "loss": 0.0574,
+      "step": 644
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8671025926443464e-06,
+      "loss": 0.0481,
+      "step": 645
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.843552548973272e-06,
+      "loss": 0.0531,
+      "step": 646
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8201368802381659e-06,
+      "loss": 0.0528,
+      "step": 647
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7968559722048906e-06,
+      "loss": 0.0534,
+      "step": 648
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7737102084191704e-06,
+      "loss": 0.0493,
+      "step": 649
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7506999702002682e-06,
+      "loss": 0.0471,
+      "step": 650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7278256366347034e-06,
+      "loss": 0.0531,
+      "step": 651
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7050875845700066e-06,
+      "loss": 0.0605,
+      "step": 652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6824861886085231e-06,
+      "loss": 0.0544,
+      "step": 653
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.660021821101222e-06,
+      "loss": 0.0506,
+      "step": 654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6376948521415714e-06,
+      "loss": 0.052,
+      "step": 655
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6155056495594467e-06,
+      "loss": 0.0433,
+      "step": 656
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5934545789150625e-06,
+      "loss": 0.0443,
+      "step": 657
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5715420034929475e-06,
+      "loss": 0.0508,
+      "step": 658
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.549768284295975e-06,
+      "loss": 0.0438,
+      "step": 659
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.528133780039397e-06,
+      "loss": 0.0499,
+      "step": 660
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5066388471449457e-06,
+      "loss": 0.0557,
+      "step": 661
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4852838397349545e-06,
+      "loss": 0.0444,
+      "step": 662
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4640691096265358e-06,
+      "loss": 0.051,
+      "step": 663
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4429950063257713e-06,
+      "loss": 0.0507,
+      "step": 664
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4220618770219608e-06,
+      "loss": 0.0582,
+      "step": 665
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.401270066581899e-06,
+      "loss": 0.0457,
+      "step": 666
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3806199175442048e-06,
+      "loss": 0.0457,
+      "step": 667
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.360111770113659e-06,
+      "loss": 0.0486,
+      "step": 668
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.0542,
+      "step": 669
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3195228291904271e-06,
+      "loss": 0.0573,
+      "step": 670
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2994427043879277e-06,
+      "loss": 0.0457,
+      "step": 671
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.279505918561923e-06,
+      "loss": 0.0517,
+      "step": 672
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2597128001647695e-06,
+      "loss": 0.0507,
+      "step": 673
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.240063675281934e-06,
+      "loss": 0.0534,
+      "step": 674
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.220558867626639e-06,
+      "loss": 0.0514,
+      "step": 675
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2011986985345313e-06,
+      "loss": 0.0524,
+      "step": 676
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1819834869583768e-06,
+      "loss": 0.0489,
+      "step": 677
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1629135494628097e-06,
+      "loss": 0.0531,
+      "step": 678
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1439892002191234e-06,
+      "loss": 0.052,
+      "step": 679
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1252107510000843e-06,
+      "loss": 0.0548,
+      "step": 680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1065785111748117e-06,
+      "loss": 0.046,
+      "step": 681
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0880927877036608e-06,
+      "loss": 0.0539,
+      "step": 682
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0697538851331769e-06,
+      "loss": 0.0534,
+      "step": 683
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.051562105591082e-06,
+      "loss": 0.0513,
+      "step": 684
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0335177487812864e-06,
+      "loss": 0.0496,
+      "step": 685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0156211119789582e-06,
+      "loss": 0.0517,
+      "step": 686
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.978724900256265e-07,
+      "loss": 0.0516,
+      "step": 687
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.802721753243193e-07,
+      "loss": 0.05,
+      "step": 688
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.628204578347466e-07,
+      "loss": 0.0476,
+      "step": 689
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.455176250685338e-07,
+      "loss": 0.0549,
+      "step": 690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.283639620844687e-07,
+      "loss": 0.0577,
+      "step": 691
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.113597514838135e-07,
+      "loss": 0.0518,
+      "step": 692
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.945052734056581e-07,
+      "loss": 0.0542,
+      "step": 693
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.778008055222809e-07,
+      "loss": 0.0538,
+      "step": 694
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.612466230346018e-07,
+      "loss": 0.0488,
+      "step": 695
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.448429986676298e-07,
+      "loss": 0.0541,
+      "step": 696
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.285902026659743e-07,
+      "loss": 0.0458,
+      "step": 697
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.124885027894014e-07,
+      "loss": 0.0508,
+      "step": 698
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.965381643084069e-07,
+      "loss": 0.0495,
+      "step": 699
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807394499998578e-07,
+      "loss": 0.0558,
+      "step": 700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.650926201426634e-07,
+      "loss": 0.0493,
+      "step": 701
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.495979325134806e-07,
+      "loss": 0.0449,
+      "step": 702
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.342556423824676e-07,
+      "loss": 0.0525,
+      "step": 703
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.19066002509089e-07,
+      "loss": 0.0457,
+      "step": 704
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.040292631379386e-07,
+      "loss": 0.0527,
+      "step": 705
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.891456719946188e-07,
+      "loss": 0.0544,
+      "step": 706
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.744154742816722e-07,
+      "loss": 0.0506,
+      "step": 707
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.598389126745209e-07,
+      "loss": 0.058,
+      "step": 708
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.45416227317488e-07,
+      "loss": 0.0497,
+      "step": 709
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.311476558198337e-07,
+      "loss": 0.0503,
+      "step": 710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.170334332518325e-07,
+      "loss": 0.0487,
+      "step": 711
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.0587,
+      "step": 712
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.892689624678327e-07,
+      "loss": 0.0439,
+      "step": 713
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.756191716628556e-07,
+      "loss": 0.0556,
+      "step": 714
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.62124644602049e-07,
+      "loss": 0.0501,
+      "step": 715
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.487856036035488e-07,
+      "loss": 0.0507,
+      "step": 716
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.35602268423906e-07,
+      "loss": 0.049,
+      "step": 717
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.225748562544741e-07,
+      "loss": 0.0506,
+      "step": 718
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.097035817178154e-07,
+      "loss": 0.0572,
+      "step": 719
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.969886568641757e-07,
+      "loss": 0.0521,
+      "step": 720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.84430291167991e-07,
+      "loss": 0.0537,
+      "step": 721
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7202869152443096e-07,
+      "loss": 0.0506,
+      "step": 722
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.597840622459937e-07,
+      "loss": 0.0601,
+      "step": 723
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4769660505914136e-07,
+      "loss": 0.0531,
+      "step": 724
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.357665191009708e-07,
+      "loss": 0.0495,
+      "step": 725
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2399400091594154e-07,
+      "loss": 0.0502,
+      "step": 726
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.123792444526287e-07,
+      "loss": 0.0496,
+      "step": 727
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0092244106053393e-07,
+      "loss": 0.0506,
+      "step": 728
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8962377948693395e-07,
+      "loss": 0.0534,
+      "step": 729
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7848344587376297e-07,
+      "loss": 0.0528,
+      "step": 730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6750162375455366e-07,
+      "loss": 0.0495,
+      "step": 731
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.566784940514145e-07,
+      "loss": 0.0506,
+      "step": 732
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.460142350720397e-07,
+      "loss": 0.0482,
+      "step": 733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.355090225067792e-07,
+      "loss": 0.0457,
+      "step": 734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2516302942574794e-07,
+      "loss": 0.0469,
+      "step": 735
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1497642627596247e-07,
+      "loss": 0.0507,
+      "step": 736
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0494938087854306e-07,
+      "loss": 0.0498,
+      "step": 737
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9508205842594727e-07,
+      "loss": 0.0551,
+      "step": 738
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8537462147924187e-07,
+      "loss": 0.0515,
+      "step": 739
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.758272299654341e-07,
+      "loss": 0.0523,
+      "step": 740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6644004117483357e-07,
+      "loss": 0.0465,
+      "step": 741
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.572132097584523e-07,
+      "loss": 0.0624,
+      "step": 742
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.481468877254722e-07,
+      "loss": 0.0511,
+      "step": 743
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.392412244407294e-07,
+      "loss": 0.0462,
+      "step": 744
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3049636662225706e-07,
+      "loss": 0.0534,
+      "step": 745
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2191245833886988e-07,
+      "loss": 0.0484,
+      "step": 746
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1348964100778914e-07,
+      "loss": 0.0528,
+      "step": 747
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.052280533923101e-07,
+      "loss": 0.0524,
+      "step": 748
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9712783159952064e-07,
+      "loss": 0.0464,
+      "step": 749
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8918910907805733e-07,
+      "loss": 0.0531,
+      "step": 750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8141201661590501e-07,
+      "loss": 0.0474,
+      "step": 751
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.737966823382442e-07,
+      "loss": 0.0524,
+      "step": 752
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6634323170533928e-07,
+      "loss": 0.0527,
+      "step": 753
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5905178751047135e-07,
+      "loss": 0.0484,
+      "step": 754
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.045,
+      "step": 755
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4495539626097289e-07,
+      "loss": 0.0511,
+      "step": 756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3815068144000353e-07,
+      "loss": 0.053,
+      "step": 757
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3150843752057442e-07,
+      "loss": 0.047,
+      "step": 758
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2502877393158587e-07,
+      "loss": 0.0497,
+      "step": 759
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1871179742348416e-07,
+      "loss": 0.0476,
+      "step": 760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1255761206649063e-07,
+      "loss": 0.0539,
+      "step": 761
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0656631924889749e-07,
+      "loss": 0.0426,
+      "step": 762
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0073801767539249e-07,
+      "loss": 0.0535,
+      "step": 763
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.507280336542912e-08,
+      "loss": 0.0458,
+      "step": 764
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.957076965165234e-08,
+      "loss": 0.0473,
+      "step": 765
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.423200717835978e-08,
+      "loss": 0.0451,
+      "step": 766
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.905660390000069e-08,
+      "loss": 0.0538,
+      "step": 767
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.404464507973608e-08,
+      "loss": 0.054,
+      "step": 768
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.919621328802973e-08,
+      "loss": 0.0501,
+      "step": 769
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.451138840128601e-08,
+      "loss": 0.0486,
+      "step": 770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.999024760054095e-08,
+      "loss": 0.0488,
+      "step": 771
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.5632865370183196e-08,
+      "loss": 0.0502,
+      "step": 772
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.143931349672837e-08,
+      "loss": 0.0461,
+      "step": 773
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.740966106764222e-08,
+      "loss": 0.0527,
+      "step": 774
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.354397447019376e-08,
+      "loss": 0.0475,
+      "step": 775
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.984231739036948e-08,
+      "loss": 0.0456,
+      "step": 776
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.630475081181861e-08,
+      "loss": 0.0513,
+      "step": 777
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.293133301485063e-08,
+      "loss": 0.0506,
+      "step": 778
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9722119575473772e-08,
+      "loss": 0.0575,
+      "step": 779
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.667716336448356e-08,
+      "loss": 0.0435,
+      "step": 780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3796514546585714e-08,
+      "loss": 0.055,
+      "step": 781
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1080220579573485e-08,
+      "loss": 0.0591,
+      "step": 782
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8528326213548276e-08,
+      "loss": 0.0563,
+      "step": 783
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.614087349017246e-08,
+      "loss": 0.0512,
+      "step": 784
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3917901741989926e-08,
+      "loss": 0.0486,
+      "step": 785
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1859447591769934e-08,
+      "loss": 0.0584,
+      "step": 786
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.965544951902052e-09,
+      "loss": 0.0454,
+      "step": 787
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.236225023844357e-09,
+      "loss": 0.0483,
+      "step": 788
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.671516297606095e-09,
+      "loss": 0.0463,
+      "step": 789
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.271444551276928e-09,
+      "loss": 0.0475,
+      "step": 790
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.036032850601723e-09,
+      "loss": 0.0514,
+      "step": 791
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9653015486064143e-09,
+      "loss": 0.0437,
+      "step": 792
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.059268285254934e-09,
+      "loss": 0.0509,
+      "step": 793
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3179479871638923e-09,
+      "loss": 0.0622,
+      "step": 794
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.413528673549941e-10,
+      "loss": 0.0522,
+      "step": 795
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.2949242505408987e-10,
+      "loss": 0.0489,
+      "step": 796
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.237344553241322e-11,
+      "loss": 0.0487,
+      "step": 797
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.054,
+      "step": 798
+    },
+    {
+      "epoch": 2.0,
+      "step": 798,
+      "total_flos": 56963522101248.0,
+      "train_loss": 0.06176260514254857,
+      "train_runtime": 711.0192,
+      "train_samples_per_second": 143.58,
+      "train_steps_per_second": 1.122
+    }
+  ],
+  "max_steps": 798,
+  "num_train_epochs": 2,
+  "total_flos": 56963522101248.0,
+  "trial_name": null,
+  "trial_params": null
+}