diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5293 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 878,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 1.0483,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 1.1094,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.0571,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 1.0518,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.7412,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.7324,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.3496,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 0.2513,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.1516,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.1852,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 0.1475,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.4109,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.1464,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.2228,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.1702,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 0.1675,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.1533,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.1614,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.1678,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.1853,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.1674,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.1614,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.1516,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.1515,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.851851851851852e-05,
+      "loss": 0.1589,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.1372,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1442,
+      "step": 27
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999931858743692e-05,
+      "loss": 0.1534,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999727435903407e-05,
+      "loss": 0.1293,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999386734265068e-05,
+      "loss": 0.1294,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999890975847185e-05,
+      "loss": 0.1412,
+      "step": 31
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99982965150241e-05,
+      "loss": 0.1292,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997547012279245e-05,
+      "loss": 0.131,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996661260451706e-05,
+      "loss": 0.1376,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999563927161272e-05,
+      "loss": 0.1613,
+      "step": 35
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994481059690223e-05,
+      "loss": 0.1293,
+      "step": 36
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993186640468603e-05,
+      "loss": 0.1483,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991756031588542e-05,
+      "loss": 0.1418,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999018925254673e-05,
+      "loss": 0.1465,
+      "step": 39
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998848632469563e-05,
+      "loss": 0.1307,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986647271243163e-05,
+      "loss": 0.131,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9984672117252425e-05,
+      "loss": 0.1281,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982560889641294e-05,
+      "loss": 0.1248,
+      "step": 43
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998031361718213e-05,
+      "loss": 0.1115,
+      "step": 44
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977930330501308e-05,
+      "loss": 0.1057,
+      "step": 45
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975411062078867e-05,
+      "loss": 0.1266,
+      "step": 46
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972755846248032e-05,
+      "loss": 0.1245,
+      "step": 47
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9969964719194745e-05,
+      "loss": 0.1028,
+      "step": 48
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967037718957193e-05,
+      "loss": 0.1313,
+      "step": 49
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963974885425267e-05,
+      "loss": 0.0989,
+      "step": 50
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996077626034003e-05,
+      "loss": 0.1124,
+      "step": 51
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957441887293157e-05,
+      "loss": 0.0865,
+      "step": 52
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995397181172631e-05,
+      "loss": 0.1292,
+      "step": 53
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950366080930557e-05,
+      "loss": 0.1418,
+      "step": 54
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9946624744045706e-05,
+      "loss": 0.1255,
+      "step": 55
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994274785205963e-05,
+      "loss": 0.1333,
+      "step": 56
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993873545780759e-05,
+      "loss": 0.1125,
+      "step": 57
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993458761597151e-05,
+      "loss": 0.1218,
+      "step": 58
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930304383079204e-05,
+      "loss": 0.129,
+      "step": 59
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992588581750366e-05,
+      "loss": 0.1005,
+      "step": 60
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921331979462198e-05,
+      "loss": 0.1092,
+      "step": 61
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991664293101566e-05,
+      "loss": 0.1031,
+      "step": 62
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9911818736067586e-05,
+      "loss": 0.1084,
+      "step": 63
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9906859460363307e-05,
+      "loss": 0.1066,
+      "step": 64
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990176517148909e-05,
+      "loss": 0.1027,
+      "step": 65
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989653593887117e-05,
+      "loss": 0.1306,
+      "step": 66
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9891171833774856e-05,
+      "loss": 0.1161,
+      "step": 67
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988567292930351e-05,
+      "loss": 0.0839,
+      "step": 68
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988003930039759e-05,
+      "loss": 0.106,
+      "step": 69
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9874271023833604e-05,
+      "loss": 0.0994,
+      "step": 70
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9868368178223078e-05,
+      "loss": 0.1191,
+      "step": 71
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986233084401147e-05,
+      "loss": 0.0963,
+      "step": 72
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9856159103477085e-05,
+      "loss": 0.1035,
+      "step": 73
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984985304072996e-05,
+      "loss": 0.1109,
+      "step": 74
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843412741710706e-05,
+      "loss": 0.077,
+      "step": 75
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9836838294189325e-05,
+      "loss": 0.1133,
+      "step": 76
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.983012978776405e-05,
+      "loss": 0.11,
+      "step": 77
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9823287313860088e-05,
+      "loss": 0.0978,
+      "step": 78
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981631096572839e-05,
+      "loss": 0.1453,
+      "step": 79
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9809200838444382e-05,
+      "loss": 0.118,
+      "step": 80
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980195702890667e-05,
+      "loss": 0.0833,
+      "step": 81
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9794579635835705e-05,
+      "loss": 0.1054,
+      "step": 82
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9787068759772458e-05,
+      "loss": 0.0974,
+      "step": 83
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9779424503077033e-05,
+      "loss": 0.136,
+      "step": 84
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977164696992728e-05,
+      "loss": 0.1076,
+      "step": 85
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9763736266317374e-05,
+      "loss": 0.1195,
+      "step": 86
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9755692500056376e-05,
+      "loss": 0.0978,
+      "step": 87
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974751578076675e-05,
+      "loss": 0.0977,
+      "step": 88
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9739206219882884e-05,
+      "loss": 0.1149,
+      "step": 89
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9730763930649556e-05,
+      "loss": 0.1064,
+      "step": 90
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9722189028120412e-05,
+      "loss": 0.0991,
+      "step": 91
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9713481629156373e-05,
+      "loss": 0.0836,
+      "step": 92
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970464185242406e-05,
+      "loss": 0.0988,
+      "step": 93
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9695669818394178e-05,
+      "loss": 0.084,
+      "step": 94
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968656564933985e-05,
+      "loss": 0.0956,
+      "step": 95
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967732946933499e-05,
+      "loss": 0.0971,
+      "step": 96
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9667961404252575e-05,
+      "loss": 0.1266,
+      "step": 97
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9658461581762948e-05,
+      "loss": 0.0883,
+      "step": 98
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964883013133208e-05,
+      "loss": 0.0886,
+      "step": 99
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9639067184219796e-05,
+      "loss": 0.0773,
+      "step": 100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9629172873477995e-05,
+      "loss": 0.0977,
+      "step": 101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9619147333948826e-05,
+      "loss": 0.0775,
+      "step": 102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960899070226286e-05,
+      "loss": 0.0986,
+      "step": 103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9598703116837232e-05,
+      "loss": 0.107,
+      "step": 104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9588284717873738e-05,
+      "loss": 0.0919,
+      "step": 105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957773564735693e-05,
+      "loss": 0.0911,
+      "step": 106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95670560490522e-05,
+      "loss": 0.0688,
+      "step": 107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9556246068503796e-05,
+      "loss": 0.1055,
+      "step": 108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954530585303285e-05,
+      "loss": 0.0968,
+      "step": 109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953423555173536e-05,
+      "loss": 0.0933,
+      "step": 110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952303531548018e-05,
+      "loss": 0.0715,
+      "step": 111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9511705296906944e-05,
+      "loss": 0.0935,
+      "step": 112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950024565042399e-05,
+      "loss": 0.0894,
+      "step": 113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948865653220626e-05,
+      "loss": 0.0881,
+      "step": 114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9476938100193166e-05,
+      "loss": 0.0903,
+      "step": 115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9465090514086448e-05,
+      "loss": 0.0924,
+      "step": 116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9453113935347984e-05,
+      "loss": 0.128,
+      "step": 117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9441008527197603e-05,
+      "loss": 0.083,
+      "step": 118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9428774454610845e-05,
+      "loss": 0.0718,
+      "step": 119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9416411884316725e-05,
+      "loss": 0.0865,
+      "step": 120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9403920984795453e-05,
+      "loss": 0.1097,
+      "step": 121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9391301926276157e-05,
+      "loss": 0.1102,
+      "step": 122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378554880734527e-05,
+      "loss": 0.107,
+      "step": 123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9365680021890508e-05,
+      "loss": 0.0851,
+      "step": 124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935267752520591e-05,
+      "loss": 0.0894,
+      "step": 125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9339547567882023e-05,
+      "loss": 0.1031,
+      "step": 126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9326290328857212e-05,
+      "loss": 0.1006,
+      "step": 127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931290598880445e-05,
+      "loss": 0.0907,
+      "step": 128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9299394730128896e-05,
+      "loss": 0.09,
+      "step": 129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928575673696537e-05,
+      "loss": 0.0969,
+      "step": 130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9271992195175875e-05,
+      "loss": 0.0983,
+      "step": 131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9258101292347042e-05,
+      "loss": 0.1079,
+      "step": 132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9244084217787587e-05,
+      "loss": 0.0899,
+      "step": 133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9229941162525727e-05,
+      "loss": 0.0891,
+      "step": 134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9215672319306566e-05,
+      "loss": 0.087,
+      "step": 135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920127788258949e-05,
+      "loss": 0.0871,
+      "step": 136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9186758048545497e-05,
+      "loss": 0.1179,
+      "step": 137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917211301505453e-05,
+      "loss": 0.092,
+      "step": 138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9157342981702792e-05,
+      "loss": 0.1039,
+      "step": 139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914244814978001e-05,
+      "loss": 0.0944,
+      "step": 140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9127428722276686e-05,
+      "loss": 0.1044,
+      "step": 141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911228490388136e-05,
+      "loss": 0.1041,
+      "step": 142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.909701690097779e-05,
+      "loss": 0.1006,
+      "step": 143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9081624921642156e-05,
+      "loss": 0.0963,
+      "step": 144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9066109175640225e-05,
+      "loss": 0.0804,
+      "step": 145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9050469874424477e-05,
+      "loss": 0.1108,
+      "step": 146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903470723113124e-05,
+      "loss": 0.1114,
+      "step": 147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9018821460577776e-05,
+      "loss": 0.0885,
+      "step": 148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9002812779259364e-05,
+      "loss": 0.0765,
+      "step": 149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8986681405346323e-05,
+      "loss": 0.0916,
+      "step": 150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8970427558681083e-05,
+      "loss": 0.081,
+      "step": 151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.895405146077514e-05,
+      "loss": 0.0744,
+      "step": 152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8937553334806077e-05,
+      "loss": 0.0775,
+      "step": 153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8920933405614497e-05,
+      "loss": 0.0741,
+      "step": 154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8904191899700978e-05,
+      "loss": 0.0738,
+      "step": 155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.888732904522296e-05,
+      "loss": 0.0867,
+      "step": 156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8870345071991662e-05,
+      "loss": 0.0958,
+      "step": 157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8853240211468945e-05,
+      "loss": 0.1035,
+      "step": 158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883601469676414e-05,
+      "loss": 0.0714,
+      "step": 159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881866876263089e-05,
+      "loss": 0.1037,
+      "step": 160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.880120264546395e-05,
+      "loss": 0.0968,
+      "step": 161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8783616583295942e-05,
+      "loss": 0.0849,
+      "step": 162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8765910815794152e-05,
+      "loss": 0.072,
+      "step": 163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874808558425722e-05,
+      "loss": 0.0917,
+      "step": 164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8730141131611882e-05,
+      "loss": 0.1045,
+      "step": 165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.871207770240965e-05,
+      "loss": 0.0881,
+      "step": 166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8693895542823477e-05,
+      "loss": 0.0858,
+      "step": 167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8675594900644395e-05,
+      "loss": 0.0916,
+      "step": 168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8657176025278163e-05,
+      "loss": 0.0789,
+      "step": 169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8638639167741836e-05,
+      "loss": 0.0927,
+      "step": 170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8619984580660365e-05,
+      "loss": 0.0942,
+      "step": 171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8601212518263157e-05,
+      "loss": 0.0872,
+      "step": 172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.858232323638059e-05,
+      "loss": 0.0931,
+      "step": 173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8563316992440545e-05,
+      "loss": 0.0833,
+      "step": 174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8544194045464888e-05,
+      "loss": 0.0958,
+      "step": 175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8524954656065944e-05,
+      "loss": 0.0795,
+      "step": 176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8505599086442956e-05,
+      "loss": 0.0809,
+      "step": 177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.848612760037848e-05,
+      "loss": 0.0982,
+      "step": 178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846654046323482e-05,
+      "loss": 0.0992,
+      "step": 179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.844683794195041e-05,
+      "loss": 0.1059,
+      "step": 180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8427020305036158e-05,
+      "loss": 0.0951,
+      "step": 181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8407087822571794e-05,
+      "loss": 0.0936,
+      "step": 182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83870407662022e-05,
+      "loss": 0.0884,
+      "step": 183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8366879409133703e-05,
+      "loss": 0.0955,
+      "step": 184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8346604026130335e-05,
+      "loss": 0.0976,
+      "step": 185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8326214893510115e-05,
+      "loss": 0.0804,
+      "step": 186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8305712289141266e-05,
+      "loss": 0.0918,
+      "step": 187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8285096492438424e-05,
+      "loss": 0.0855,
+      "step": 188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8264367784358856e-05,
+      "loss": 0.0954,
+      "step": 189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8243526447398595e-05,
+      "loss": 0.1073,
+      "step": 190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8222572765588626e-05,
+      "loss": 0.0913,
+      "step": 191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8201507024490986e-05,
+      "loss": 0.0831,
+      "step": 192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818032951119489e-05,
+      "loss": 0.0828,
+      "step": 193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8159040514312805e-05,
+      "loss": 0.0937,
+      "step": 194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8137640323976536e-05,
+      "loss": 0.0801,
+      "step": 195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8116129231833247e-05,
+      "loss": 0.0879,
+      "step": 196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8094507531041516e-05,
+      "loss": 0.0911,
+      "step": 197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8072775516267306e-05,
+      "loss": 0.0628,
+      "step": 198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8050933483679974e-05,
+      "loss": 0.0794,
+      "step": 199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8028981730948238e-05,
+      "loss": 0.0734,
+      "step": 200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800692055723609e-05,
+      "loss": 0.0789,
+      "step": 201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7984750263198752e-05,
+      "loss": 0.0857,
+      "step": 202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7962471150978565e-05,
+      "loss": 0.0966,
+      "step": 203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.794008352420086e-05,
+      "loss": 0.0948,
+      "step": 204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7917587687969847e-05,
+      "loss": 0.0984,
+      "step": 205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789498394886443e-05,
+      "loss": 0.089,
+      "step": 206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7872272614934053e-05,
+      "loss": 0.0636,
+      "step": 207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.784945399569447e-05,
+      "loss": 0.0777,
+      "step": 208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7826528402123565e-05,
+      "loss": 0.1028,
+      "step": 209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7803496146657086e-05,
+      "loss": 0.1008,
+      "step": 210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7780357543184396e-05,
+      "loss": 0.0878,
+      "step": 211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.1061,
+      "step": 212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7733762555020235e-05,
+      "loss": 0.0742,
+      "step": 213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7710306805336973e-05,
+      "loss": 0.0965,
+      "step": 214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7686745977655254e-05,
+      "loss": 0.1249,
+      "step": 215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766308039306797e-05,
+      "loss": 0.0999,
+      "step": 216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7639310374095645e-05,
+      "loss": 0.0792,
+      "step": 217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7615436244682068e-05,
+      "loss": 0.0806,
+      "step": 218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.759145833018988e-05,
+      "loss": 0.0883,
+      "step": 219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7567376957396124e-05,
+      "loss": 0.1087,
+      "step": 220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7543192454487793e-05,
+      "loss": 0.0889,
+      "step": 221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.751890515105738e-05,
+      "loss": 0.0798,
+      "step": 222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7494515378098348e-05,
+      "loss": 0.0809,
+      "step": 223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7470023468000655e-05,
+      "loss": 0.0947,
+      "step": 224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.744542975454621e-05,
+      "loss": 0.0944,
+      "step": 225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.742073457290431e-05,
+      "loss": 0.1034,
+      "step": 226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7395938259627102e-05,
+      "loss": 0.1006,
+      "step": 227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7371041152644975e-05,
+      "loss": 0.0778,
+      "step": 228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7346043591261958e-05,
+      "loss": 0.0839,
+      "step": 229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7320945916151092e-05,
+      "loss": 0.0854,
+      "step": 230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7295748469349805e-05,
+      "loss": 0.0817,
+      "step": 231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7270451594255232e-05,
+      "loss": 0.0867,
+      "step": 232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7245055635619543e-05,
+      "loss": 0.1,
+      "step": 233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7219560939545246e-05,
+      "loss": 0.0801,
+      "step": 234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.719396785348046e-05,
+      "loss": 0.0757,
+      "step": 235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7168276726214198e-05,
+      "loss": 0.0685,
+      "step": 236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7142487907871595e-05,
+      "loss": 0.0808,
+      "step": 237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.711660174990914e-05,
+      "loss": 0.0707,
+      "step": 238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7090618605109913e-05,
+      "loss": 0.0817,
+      "step": 239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7064538827578724e-05,
+      "loss": 0.0865,
+      "step": 240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.703836277273735e-05,
+      "loss": 0.0852,
+      "step": 241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701209079731963e-05,
+      "loss": 0.1098,
+      "step": 242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698572325936665e-05,
+      "loss": 0.0771,
+      "step": 243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6959260518221844e-05,
+      "loss": 0.0623,
+      "step": 244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.69327029345261e-05,
+      "loss": 0.1037,
+      "step": 245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6906050870212834e-05,
+      "loss": 0.0766,
+      "step": 246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6879304688503088e-05,
+      "loss": 0.0987,
+      "step": 247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685246475390053e-05,
+      "loss": 0.1071,
+      "step": 248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6825531432186545e-05,
+      "loss": 0.0783,
+      "step": 249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.67985050904152e-05,
+      "loss": 0.0962,
+      "step": 250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.677138609690828e-05,
+      "loss": 0.0758,
+      "step": 251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6744174821250237e-05,
+      "loss": 0.0753,
+      "step": 252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6716871634283173e-05,
+      "loss": 0.0823,
+      "step": 253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6689476908101784e-05,
+      "loss": 0.0934,
+      "step": 254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6661991016048275e-05,
+      "loss": 0.0826,
+      "step": 255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6634414332707297e-05,
+      "loss": 0.099,
+      "step": 256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6606747233900816e-05,
+      "loss": 0.0823,
+      "step": 257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6578990096683005e-05,
+      "loss": 0.0908,
+      "step": 258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6551143299335107e-05,
+      "loss": 0.0739,
+      "step": 259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6523207221360267e-05,
+      "loss": 0.0919,
+      "step": 260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6495182243478382e-05,
+      "loss": 0.085,
+      "step": 261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6467068747620888e-05,
+      "loss": 0.1037,
+      "step": 262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6438867116925572e-05,
+      "loss": 0.0875,
+      "step": 263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6410577735731346e-05,
+      "loss": 0.0957,
+      "step": 264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6382200989572997e-05,
+      "loss": 0.0914,
+      "step": 265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6353737265175963e-05,
+      "loss": 0.0894,
+      "step": 266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.632518695045102e-05,
+      "loss": 0.0815,
+      "step": 267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629655043448904e-05,
+      "loss": 0.0806,
+      "step": 268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.626782810755565e-05,
+      "loss": 0.0714,
+      "step": 269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6239020361085947e-05,
+      "loss": 0.0834,
+      "step": 270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.621012758767913e-05,
+      "loss": 0.0934,
+      "step": 271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618115018109318e-05,
+      "loss": 0.0778,
+      "step": 272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.615208853623947e-05,
+      "loss": 0.1026,
+      "step": 273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6122943049177403e-05,
+      "loss": 0.098,
+      "step": 274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6093714117109e-05,
+      "loss": 0.0744,
+      "step": 275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6064402138373488e-05,
+      "loss": 0.0823,
+      "step": 276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6035007512441883e-05,
+      "loss": 0.0849,
+      "step": 277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6005530639911525e-05,
+      "loss": 0.0755,
+      "step": 278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5975971922500643e-05,
+      "loss": 0.0823,
+      "step": 279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.594633176304287e-05,
+      "loss": 0.0976,
+      "step": 280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5916610565481737e-05,
+      "loss": 0.0892,
+      "step": 281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5886808734865202e-05,
+      "loss": 0.0672,
+      "step": 282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5856926677340093e-05,
+      "loss": 0.0934,
+      "step": 283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.58269648001466e-05,
+      "loss": 0.0974,
+      "step": 284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.579692351161272e-05,
+      "loss": 0.0892,
+      "step": 285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5766803221148676e-05,
+      "loss": 0.0789,
+      "step": 286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.573660433924135e-05,
+      "loss": 0.1059,
+      "step": 287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.57063272774487e-05,
+      "loss": 0.0829,
+      "step": 288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5675972448394126e-05,
+      "loss": 0.0701,
+      "step": 289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5645540265760874e-05,
+      "loss": 0.0895,
+      "step": 290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5615031144286364e-05,
+      "loss": 0.0776,
+      "step": 291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5584445499756577e-05,
+      "loss": 0.0799,
+      "step": 292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5553783749000363e-05,
+      "loss": 0.0776,
+      "step": 293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5523046309883757e-05,
+      "loss": 0.0933,
+      "step": 294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5492233601304313e-05,
+      "loss": 0.1033,
+      "step": 295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5461346043185358e-05,
+      "loss": 0.0979,
+      "step": 296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5430384056470294e-05,
+      "loss": 0.0882,
+      "step": 297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5399348063116858e-05,
+      "loss": 0.0879,
+      "step": 298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5368238486091372e-05,
+      "loss": 0.0831,
+      "step": 299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5337055749362957e-05,
+      "loss": 0.1077,
+      "step": 300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.530580027789779e-05,
+      "loss": 0.0758,
+      "step": 301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527447249765329e-05,
+      "loss": 0.0956,
+      "step": 302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5243072835572319e-05,
+      "loss": 0.0901,
+      "step": 303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5211601719577358e-05,
+      "loss": 0.0794,
+      "step": 304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.518005957856468e-05,
+      "loss": 0.0773,
+      "step": 305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5148446842398518e-05,
+      "loss": 0.0758,
+      "step": 306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5116763941905177e-05,
+      "loss": 0.0837,
+      "step": 307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5085011308867183e-05,
+      "loss": 0.0881,
+      "step": 308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5053189376017407e-05,
+      "loss": 0.1084,
+      "step": 309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5021298577033134e-05,
+      "loss": 0.0818,
+      "step": 310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4989339346530197e-05,
+      "loss": 0.0742,
+      "step": 311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4957312120057006e-05,
+      "loss": 0.0985,
+      "step": 312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4925217334088662e-05,
+      "loss": 0.0738,
+      "step": 313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4893055426020969e-05,
+      "loss": 0.0979,
+      "step": 314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4860826834164489e-05,
+      "loss": 0.0658,
+      "step": 315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4828531997738574e-05,
+      "loss": 0.085,
+      "step": 316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4796171356865362e-05,
+      "loss": 0.101,
+      "step": 317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4763745352563805e-05,
+      "loss": 0.0761,
+      "step": 318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4731254426743633e-05,
+      "loss": 0.0946,
+      "step": 319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.469869902219935e-05,
+      "loss": 0.0781,
+      "step": 320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4666079582604185e-05,
+      "loss": 0.08,
+      "step": 321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4633396552504062e-05,
+      "loss": 0.0802,
+      "step": 322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4600650377311523e-05,
+      "loss": 0.0614,
+      "step": 323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4567841503299673e-05,
+      "loss": 0.0828,
+      "step": 324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4534970377596089e-05,
+      "loss": 0.0691,
+      "step": 325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4502037448176734e-05,
+      "loss": 0.096,
+      "step": 326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4469043163859844e-05,
+      "loss": 0.0827,
+      "step": 327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4435987974299815e-05,
+      "loss": 0.099,
+      "step": 328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4402872329981077e-05,
+      "loss": 0.0915,
+      "step": 329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4369696682211948e-05,
+      "loss": 0.075,
+      "step": 330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4336461483118498e-05,
+      "loss": 0.0966,
+      "step": 331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4303167185638367e-05,
+      "loss": 0.0933,
+      "step": 332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4269814243514608e-05,
+      "loss": 0.0839,
+      "step": 333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4236403111289494e-05,
+      "loss": 0.0893,
+      "step": 334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.420293424429833e-05,
+      "loss": 0.0718,
+      "step": 335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4169408098663249e-05,
+      "loss": 0.088,
+      "step": 336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4135825131286984e-05,
+      "loss": 0.0748,
+      "step": 337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4102185799846652e-05,
+      "loss": 0.1025,
+      "step": 338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.406849056278752e-05,
+      "loss": 0.0839,
+      "step": 339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4034739879316737e-05,
+      "loss": 0.096,
+      "step": 340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4000934209397103e-05,
+      "loss": 0.0804,
+      "step": 341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.396707401374078e-05,
+      "loss": 0.0975,
+      "step": 342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3933159753803021e-05,
+      "loss": 0.0751,
+      "step": 343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3899191891775885e-05,
+      "loss": 0.0896,
+      "step": 344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3865170890581925e-05,
+      "loss": 0.0688,
+      "step": 345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3831097213867902e-05,
+      "loss": 0.0905,
+      "step": 346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3796971325998434e-05,
+      "loss": 0.0926,
+      "step": 347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3762793692049702e-05,
+      "loss": 0.0599,
+      "step": 348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3728564777803089e-05,
+      "loss": 0.0748,
+      "step": 349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3694285049738833e-05,
+      "loss": 0.0863,
+      "step": 350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.365995497502969e-05,
+      "loss": 0.0826,
+      "step": 351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3625575021534536e-05,
+      "loss": 0.0958,
+      "step": 352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3591145657792018e-05,
+      "loss": 0.0793,
+      "step": 353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3556667353014159e-05,
+      "loss": 0.0927,
+      "step": 354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3522140577079954e-05,
+      "loss": 0.0965,
+      "step": 355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.348756580052899e-05,
+      "loss": 0.0989,
+      "step": 356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3452943494554998e-05,
+      "loss": 0.0657,
+      "step": 357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3418274130999468e-05,
+      "loss": 0.0935,
+      "step": 358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3383558182345203e-05,
+      "loss": 0.0825,
+      "step": 359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3348796121709862e-05,
+      "loss": 0.1046,
+      "step": 360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3313988422839552e-05,
+      "loss": 0.0861,
+      "step": 361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3279135560102336e-05,
+      "loss": 0.0836,
+      "step": 362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3244238008481785e-05,
+      "loss": 0.0711,
+      "step": 363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.32092962435705e-05,
+      "loss": 0.0832,
+      "step": 364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.317431074156363e-05,
+      "loss": 0.087,
+      "step": 365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3139281979252394e-05,
+      "loss": 0.0722,
+      "step": 366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.310421043401756e-05,
+      "loss": 0.0869,
+      "step": 367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306909658382296e-05,
+      "loss": 0.0848,
+      "step": 368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303394090720897e-05,
+      "loss": 0.0918,
+      "step": 369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299874388328598e-05,
+      "loss": 0.0754,
+      "step": 370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2963505991727878e-05,
+      "loss": 0.0869,
+      "step": 371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2928227712765504e-05,
+      "loss": 0.0808,
+      "step": 372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.289290952718011e-05,
+      "loss": 0.0999,
+      "step": 373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.28575519162968e-05,
+      "loss": 0.082,
+      "step": 374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2822155361977977e-05,
+      "loss": 0.0707,
+      "step": 375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2786720346616784e-05,
+      "loss": 0.0855,
+      "step": 376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2751247353130507e-05,
+      "loss": 0.0859,
+      "step": 377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2715736864954017e-05,
+      "loss": 0.0719,
+      "step": 378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2680189366033173e-05,
+      "loss": 0.0962,
+      "step": 379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2644605340818217e-05,
+      "loss": 0.1035,
+      "step": 380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2608985274257186e-05,
+      "loss": 0.0798,
+      "step": 381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.25733296517893e-05,
+      "loss": 0.0716,
+      "step": 382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2537638959338336e-05,
+      "loss": 0.0792,
+      "step": 383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2501913683306026e-05,
+      "loss": 0.0818,
+      "step": 384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.24661543105654e-05,
+      "loss": 0.0858,
+      "step": 385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.243036132845418e-05,
+      "loss": 0.1028,
+      "step": 386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2394535224768118e-05,
+      "loss": 0.0964,
+      "step": 387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2358676487754355e-05,
+      "loss": 0.0808,
+      "step": 388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2322785606104772e-05,
+      "loss": 0.0771,
+      "step": 389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2286863068949322e-05,
+      "loss": 0.0878,
+      "step": 390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2250909365849366e-05,
+      "loss": 0.0921,
+      "step": 391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2214924986791003e-05,
+      "loss": 0.0812,
+      "step": 392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2178910422178394e-05,
+      "loss": 0.0884,
+      "step": 393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2142866162827074e-05,
+      "loss": 0.0848,
+      "step": 394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2106792699957264e-05,
+      "loss": 0.076,
+      "step": 395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2070690525187186e-05,
+      "loss": 0.0828,
+      "step": 396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2034560130526341e-05,
+      "loss": 0.1015,
+      "step": 397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1998402008368828e-05,
+      "loss": 0.0732,
+      "step": 398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1962216651486624e-05,
+      "loss": 0.075,
+      "step": 399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1926004553022861e-05,
+      "loss": 0.0971,
+      "step": 400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1889766206485121e-05,
+      "loss": 0.0821,
+      "step": 401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1853502105738692e-05,
+      "loss": 0.0892,
+      "step": 402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1817212744999848e-05,
+      "loss": 0.0825,
+      "step": 403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1780898618829122e-05,
+      "loss": 0.0784,
+      "step": 404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1744560222124541e-05,
+      "loss": 0.0822,
+      "step": 405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1708198050114916e-05,
+      "loss": 0.0876,
+      "step": 406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1671812598353058e-05,
+      "loss": 0.0811,
+      "step": 407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1635404362709044e-05,
+      "loss": 0.0875,
+      "step": 408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1598973839363462e-05,
+      "loss": 0.0957,
+      "step": 409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1562521524800632e-05,
+      "loss": 0.0971,
+      "step": 410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1526047915801863e-05,
+      "loss": 0.0793,
+      "step": 411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1489553509438658e-05,
+      "loss": 0.0807,
+      "step": 412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1453038803065958e-05,
+      "loss": 0.1094,
+      "step": 413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1416504294315358e-05,
+      "loss": 0.0804,
+      "step": 414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1379950481088318e-05,
+      "loss": 0.1068,
+      "step": 415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1343377861549394e-05,
+      "loss": 0.064,
+      "step": 416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1306786934119433e-05,
+      "loss": 0.0764,
+      "step": 417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1270178197468788e-05,
+      "loss": 0.08,
+      "step": 418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1233552150510523e-05,
+      "loss": 0.0897,
+      "step": 419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.11969092923936e-05,
+      "loss": 0.0897,
+      "step": 420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1160250122496106e-05,
+      "loss": 0.0984,
+      "step": 421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1123575140418415e-05,
+      "loss": 0.0757,
+      "step": 422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1086884845976396e-05,
+      "loss": 0.0913,
+      "step": 423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1050179739194608e-05,
+      "loss": 0.0847,
+      "step": 424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1013460320299459e-05,
+      "loss": 0.0979,
+      "step": 425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0976727089712424e-05,
+      "loss": 0.082,
+      "step": 426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0939980548043194e-05,
+      "loss": 0.062,
+      "step": 427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0903221196082874e-05,
+      "loss": 0.0829,
+      "step": 428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.086644953479715e-05,
+      "loss": 0.0668,
+      "step": 429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0829666065319457e-05,
+      "loss": 0.0819,
+      "step": 430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0792871288944165e-05,
+      "loss": 0.0772,
+      "step": 431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.075606570711973e-05,
+      "loss": 0.0773,
+      "step": 432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0719249821441862e-05,
+      "loss": 0.0579,
+      "step": 433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0682424133646712e-05,
+      "loss": 0.0722,
+      "step": 434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0645589145603998e-05,
+      "loss": 0.0613,
+      "step": 435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.06087453593102e-05,
+      "loss": 0.0829,
+      "step": 436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0571893276881688e-05,
+      "loss": 0.0907,
+      "step": 437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.05350334005479e-05,
+      "loss": 0.0836,
+      "step": 438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.04981662326445e-05,
+      "loss": 0.0881,
+      "step": 439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0461292275606512e-05,
+      "loss": 0.0528,
+      "step": 440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0424412031961485e-05,
+      "loss": 0.0715,
+      "step": 441
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.038752600432265e-05,
+      "loss": 0.0804,
+      "step": 442
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0350634695382054e-05,
+      "loss": 0.0499,
+      "step": 443
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.031373860790373e-05,
+      "loss": 0.082,
+      "step": 444
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0276838244716824e-05,
+      "loss": 0.059,
+      "step": 445
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0239934108708762e-05,
+      "loss": 0.076,
+      "step": 446
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0203026702818383e-05,
+      "loss": 0.0625,
+      "step": 447
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0166116530029082e-05,
+      "loss": 0.0705,
+      "step": 448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0129204093361971e-05,
+      "loss": 0.0756,
+      "step": 449
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0092289895869011e-05,
+      "loss": 0.0759,
+      "step": 450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0055374440626162e-05,
+      "loss": 0.0608,
+      "step": 451
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0018458230726524e-05,
+      "loss": 0.0607,
+      "step": 452
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.981541769273478e-06,
+      "loss": 0.0788,
+      "step": 453
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.944625559373841e-06,
+      "loss": 0.0733,
+      "step": 454
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.90771010413099e-06,
+      "loss": 0.0773,
+      "step": 455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.870795906638032e-06,
+      "loss": 0.0735,
+      "step": 456
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.833883469970924e-06,
+      "loss": 0.0592,
+      "step": 457
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.79697329718162e-06,
+      "loss": 0.0724,
+      "step": 458
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.760065891291241e-06,
+      "loss": 0.0886,
+      "step": 459
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.723161755283177e-06,
+      "loss": 0.0515,
+      "step": 460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.686261392096275e-06,
+      "loss": 0.0724,
+      "step": 461
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.649365304617953e-06,
+      "loss": 0.0601,
+      "step": 462
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.612473995677354e-06,
+      "loss": 0.0472,
+      "step": 463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.57558796803852e-06,
+      "loss": 0.0765,
+      "step": 464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.538707724393491e-06,
+      "loss": 0.0739,
+      "step": 465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.501833767355502e-06,
+      "loss": 0.0851,
+      "step": 466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.464966599452103e-06,
+      "loss": 0.0604,
+      "step": 467
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.428106723118316e-06,
+      "loss": 0.0801,
+      "step": 468
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.391254640689806e-06,
+      "loss": 0.0634,
+      "step": 469
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.354410854396004e-06,
+      "loss": 0.0566,
+      "step": 470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.317575866353293e-06,
+      "loss": 0.0533,
+      "step": 471
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.28075017855814e-06,
+      "loss": 0.0581,
+      "step": 472
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.243934292880274e-06,
+      "loss": 0.063,
+      "step": 473
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.20712871105584e-06,
+      "loss": 0.0703,
+      "step": 474
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.170333934680545e-06,
+      "loss": 0.062,
+      "step": 475
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.133550465202855e-06,
+      "loss": 0.0736,
+      "step": 476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.096778803917126e-06,
+      "loss": 0.0859,
+      "step": 477
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.06001945195681e-06,
+      "loss": 0.0662,
+      "step": 478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.023272910287581e-06,
+      "loss": 0.0729,
+      "step": 479
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.986539679700543e-06,
+      "loss": 0.0791,
+      "step": 480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.949820260805397e-06,
+      "loss": 0.0714,
+      "step": 481
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.913115154023604e-06,
+      "loss": 0.0592,
+      "step": 482
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.876424859581589e-06,
+      "loss": 0.0937,
+      "step": 483
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.839749877503899e-06,
+      "loss": 0.0623,
+      "step": 484
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.803090707606403e-06,
+      "loss": 0.0705,
+      "step": 485
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.766447849489483e-06,
+      "loss": 0.075,
+      "step": 486
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.729821802531213e-06,
+      "loss": 0.0915,
+      "step": 487
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.693213065880568e-06,
+      "loss": 0.0802,
+      "step": 488
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.656622138450606e-06,
+      "loss": 0.0728,
+      "step": 489
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.620049518911683e-06,
+      "loss": 0.0633,
+      "step": 490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.583495705684647e-06,
+      "loss": 0.0593,
+      "step": 491
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.546961196934043e-06,
+      "loss": 0.0638,
+      "step": 492
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.510446490561344e-06,
+      "loss": 0.0513,
+      "step": 493
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.473952084198139e-06,
+      "loss": 0.078,
+      "step": 494
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.43747847519937e-06,
+      "loss": 0.1057,
+      "step": 495
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.401026160636543e-06,
+      "loss": 0.0675,
+      "step": 496
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.364595637290958e-06,
+      "loss": 0.0726,
+      "step": 497
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.328187401646947e-06,
+      "loss": 0.0535,
+      "step": 498
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.291801949885085e-06,
+      "loss": 0.0624,
+      "step": 499
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.25543977787546e-06,
+      "loss": 0.0648,
+      "step": 500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.21910138117088e-06,
+      "loss": 0.0809,
+      "step": 501
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.182787255000155e-06,
+      "loss": 0.075,
+      "step": 502
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.146497894261313e-06,
+      "loss": 0.0865,
+      "step": 503
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.11023379351488e-06,
+      "loss": 0.0691,
+      "step": 504
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.07399544697714e-06,
+      "loss": 0.0753,
+      "step": 505
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.037783348513376e-06,
+      "loss": 0.0688,
+      "step": 506
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.001597991631175e-06,
+      "loss": 0.0587,
+      "step": 507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.965439869473664e-06,
+      "loss": 0.0897,
+      "step": 508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.929309474812818e-06,
+      "loss": 0.0728,
+      "step": 509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.89320730004274e-06,
+      "loss": 0.0534,
+      "step": 510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.857133837172928e-06,
+      "loss": 0.0494,
+      "step": 511
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.82108957782161e-06,
+      "loss": 0.0919,
+      "step": 512
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.785075013209e-06,
+      "loss": 0.0727,
+      "step": 513
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.749090634150636e-06,
+      "loss": 0.0627,
+      "step": 514
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.71313693105068e-06,
+      "loss": 0.0779,
+      "step": 515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.677214393895228e-06,
+      "loss": 0.0614,
+      "step": 516
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.641323512245647e-06,
+      "loss": 0.055,
+      "step": 517
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.605464775231885e-06,
+      "loss": 0.0726,
+      "step": 518
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.569638671545823e-06,
+      "loss": 0.0682,
+      "step": 519
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.533845689434602e-06,
+      "loss": 0.0595,
+      "step": 520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.498086316693977e-06,
+      "loss": 0.0694,
+      "step": 521
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4623610406616674e-06,
+      "loss": 0.0634,
+      "step": 522
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.426670348210703e-06,
+      "loss": 0.0587,
+      "step": 523
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3910147257428165e-06,
+      "loss": 0.0554,
+      "step": 524
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.355394659181787e-06,
+      "loss": 0.0716,
+      "step": 525
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3198106339668285e-06,
+      "loss": 0.0594,
+      "step": 526
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.284263135045986e-06,
+      "loss": 0.0509,
+      "step": 527
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.248752646869495e-06,
+      "loss": 0.0754,
+      "step": 528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.21327965338322e-06,
+      "loss": 0.0573,
+      "step": 529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.177844638022026e-06,
+      "loss": 0.0649,
+      "step": 530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.142448083703204e-06,
+      "loss": 0.0698,
+      "step": 531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.107090472819895e-06,
+      "loss": 0.098,
+      "step": 532
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.071772287234497e-06,
+      "loss": 0.0607,
+      "step": 533
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.036494008272124e-06,
+      "loss": 0.0696,
+      "step": 534
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0012561167140215e-06,
+      "loss": 0.0708,
+      "step": 535
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.966059092791033e-06,
+      "loss": 0.0682,
+      "step": 536
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.930903416177044e-06,
+      "loss": 0.079,
+      "step": 537
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8957895659824424e-06,
+      "loss": 0.0703,
+      "step": 538
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.86071802074761e-06,
+      "loss": 0.0465,
+      "step": 539
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.82568925843637e-06,
+      "loss": 0.0788,
+      "step": 540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.790703756429503e-06,
+      "loss": 0.0778,
+      "step": 541
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7557619915182195e-06,
+      "loss": 0.0771,
+      "step": 542
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.720864439897667e-06,
+      "loss": 0.0782,
+      "step": 543
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.686011577160451e-06,
+      "loss": 0.0669,
+      "step": 544
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.651203878290139e-06,
+      "loss": 0.0691,
+      "step": 545
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.616441817654802e-06,
+      "loss": 0.0702,
+      "step": 546
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.581725869000536e-06,
+      "loss": 0.0726,
+      "step": 547
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.547056505445006e-06,
+      "loss": 0.0873,
+      "step": 548
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.512434199471016e-06,
+      "loss": 0.0652,
+      "step": 549
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.477859422920046e-06,
+      "loss": 0.0928,
+      "step": 550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4433326469858445e-06,
+      "loss": 0.0459,
+      "step": 551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.408854342207983e-06,
+      "loss": 0.0686,
+      "step": 552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.374424978465467e-06,
+      "loss": 0.0862,
+      "step": 553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.340045024970316e-06,
+      "loss": 0.0556,
+      "step": 554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.305714950261168e-06,
+      "loss": 0.0663,
+      "step": 555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2714352221969155e-06,
+      "loss": 0.0755,
+      "step": 556
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.237206307950298e-06,
+      "loss": 0.0681,
+      "step": 557
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.203028674001568e-06,
+      "loss": 0.0667,
+      "step": 558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.168902786132105e-06,
+      "loss": 0.0727,
+      "step": 559
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1348291094180766e-06,
+      "loss": 0.0585,
+      "step": 560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.10080810822412e-06,
+      "loss": 0.0648,
+      "step": 561
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0668402461969815e-06,
+      "loss": 0.0734,
+      "step": 562
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.032925986259224e-06,
+      "loss": 0.0794,
+      "step": 563
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9990657906029025e-06,
+      "loss": 0.0648,
+      "step": 564
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.965260120683265e-06,
+      "loss": 0.0753,
+      "step": 565
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.931509437212483e-06,
+      "loss": 0.0574,
+      "step": 566
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.897814200153349e-06,
+      "loss": 0.0675,
+      "step": 567
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.864174868713018e-06,
+      "loss": 0.0662,
+      "step": 568
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.830591901336753e-06,
+      "loss": 0.0687,
+      "step": 569
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7970657557016706e-06,
+      "loss": 0.0645,
+      "step": 570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.763596888710513e-06,
+      "loss": 0.0549,
+      "step": 571
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.730185756485396e-06,
+      "loss": 0.0501,
+      "step": 572
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.696832814361635e-06,
+      "loss": 0.0668,
+      "step": 573
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.663538516881503e-06,
+      "loss": 0.0691,
+      "step": 574
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6303033177880525e-06,
+      "loss": 0.0666,
+      "step": 575
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.597127670018927e-06,
+      "loss": 0.0584,
+      "step": 576
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.564012025700185e-06,
+      "loss": 0.0654,
+      "step": 577
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.53095683614016e-06,
+      "loss": 0.0691,
+      "step": 578
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.497962551823266e-06,
+      "loss": 0.0448,
+      "step": 579
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.465029622403912e-06,
+      "loss": 0.0449,
+      "step": 580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.432158496700329e-06,
+      "loss": 0.0843,
+      "step": 581
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.399349622688479e-06,
+      "loss": 0.0516,
+      "step": 582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.366603447495942e-06,
+      "loss": 0.0699,
+      "step": 583
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.333920417395817e-06,
+      "loss": 0.0705,
+      "step": 584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3013009778006545e-06,
+      "loss": 0.0778,
+      "step": 585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2687455732563665e-06,
+      "loss": 0.0696,
+      "step": 586
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.236254647436196e-06,
+      "loss": 0.0716,
+      "step": 587
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.203828643134643e-06,
+      "loss": 0.0655,
+      "step": 588
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.171468002261431e-06,
+      "loss": 0.0701,
+      "step": 589
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.139173165835514e-06,
+      "loss": 0.0803,
+      "step": 590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.106944573979034e-06,
+      "loss": 0.0679,
+      "step": 591
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.074782665911341e-06,
+      "loss": 0.0728,
+      "step": 592
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.042687879942996e-06,
+      "loss": 0.0657,
+      "step": 593
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.01066065346981e-06,
+      "loss": 0.0771,
+      "step": 594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.978701422966868e-06,
+      "loss": 0.0623,
+      "step": 595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.946810623982595e-06,
+      "loss": 0.0724,
+      "step": 596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.914988691132816e-06,
+      "loss": 0.0667,
+      "step": 597
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.883236058094825e-06,
+      "loss": 0.0659,
+      "step": 598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.851553157601484e-06,
+      "loss": 0.0614,
+      "step": 599
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.819940421435321e-06,
+      "loss": 0.061,
+      "step": 600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.788398280422646e-06,
+      "loss": 0.0519,
+      "step": 601
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.756927164427685e-06,
+      "loss": 0.0564,
+      "step": 602
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.725527502346708e-06,
+      "loss": 0.0616,
+      "step": 603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.694199722102213e-06,
+      "loss": 0.0681,
+      "step": 604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6629442506370485e-06,
+      "loss": 0.0475,
+      "step": 605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.631761513908632e-06,
+      "loss": 0.0596,
+      "step": 606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.600651936883143e-06,
+      "loss": 0.0794,
+      "step": 607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.569615943529709e-06,
+      "loss": 0.0667,
+      "step": 608
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.538653956814647e-06,
+      "loss": 0.0638,
+      "step": 609
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.507766398695691e-06,
+      "loss": 0.0596,
+      "step": 610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.476953690116245e-06,
+      "loss": 0.0475,
+      "step": 611
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.446216250999641e-06,
+      "loss": 0.0672,
+      "step": 612
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4155545002434206e-06,
+      "loss": 0.0698,
+      "step": 613
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3849688557136385e-06,
+      "loss": 0.0651,
+      "step": 614
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.354459734239128e-06,
+      "loss": 0.0746,
+      "step": 615
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3240275516058735e-06,
+      "loss": 0.0598,
+      "step": 616
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.293672722551303e-06,
+      "loss": 0.0622,
+      "step": 617
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.263395660758653e-06,
+      "loss": 0.0621,
+      "step": 618
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2331967788513295e-06,
+      "loss": 0.058,
+      "step": 619
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.20307648838728e-06,
+      "loss": 0.0665,
+      "step": 620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.173035199853401e-06,
+      "loss": 0.062,
+      "step": 621
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.143073322659912e-06,
+      "loss": 0.0764,
+      "step": 622
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.113191265134801e-06,
+      "loss": 0.0676,
+      "step": 623
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.083389434518268e-06,
+      "loss": 0.0636,
+      "step": 624
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.053668236957135e-06,
+      "loss": 0.0638,
+      "step": 625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.024028077499359e-06,
+      "loss": 0.0721,
+      "step": 626
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.994469360088479e-06,
+      "loss": 0.0715,
+      "step": 627
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.964992487558122e-06,
+      "loss": 0.0537,
+      "step": 628
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9355978616265145e-06,
+      "loss": 0.0717,
+      "step": 629
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.906285882890999e-06,
+      "loss": 0.0604,
+      "step": 630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.877056950822598e-06,
+      "loss": 0.0581,
+      "step": 631
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.847911463760529e-06,
+      "loss": 0.0448,
+      "step": 632
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8188498189068215e-06,
+      "loss": 0.0571,
+      "step": 633
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7898724123208754e-06,
+      "loss": 0.0567,
+      "step": 634
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7609796389140562e-06,
+      "loss": 0.0614,
+      "step": 635
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7321718924443507e-06,
+      "loss": 0.0542,
+      "step": 636
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.703449565510964e-06,
+      "loss": 0.0551,
+      "step": 637
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.674813049548982e-06,
+      "loss": 0.0792,
+      "step": 638
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.646262734824042e-06,
+      "loss": 0.0762,
+      "step": 639
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.617799010427001e-06,
+      "loss": 0.0574,
+      "step": 640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5894222642686596e-06,
+      "loss": 0.0699,
+      "step": 641
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5611328830744276e-06,
+      "loss": 0.0635,
+      "step": 642
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5329312523791137e-06,
+      "loss": 0.0741,
+      "step": 643
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5048177565216236e-06,
+      "loss": 0.0573,
+      "step": 644
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4767927786397358e-06,
+      "loss": 0.0421,
+      "step": 645
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4488567006648986e-06,
+      "loss": 0.0737,
+      "step": 646
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4210099033169987e-06,
+      "loss": 0.063,
+      "step": 647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3932527660991877e-06,
+      "loss": 0.0712,
+      "step": 648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.365585667292702e-06,
+      "loss": 0.0602,
+      "step": 649
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.338008983951724e-06,
+      "loss": 0.0558,
+      "step": 650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.310523091898221e-06,
+      "loss": 0.0555,
+      "step": 651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2831283657168277e-06,
+      "loss": 0.0691,
+      "step": 652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2558251787497663e-06,
+      "loss": 0.0624,
+      "step": 653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.228613903091723e-06,
+      "loss": 0.0582,
+      "step": 654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2014949095848024e-06,
+      "loss": 0.062,
+      "step": 655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.174468567813461e-06,
+      "loss": 0.0627,
+      "step": 656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1475352460994744e-06,
+      "loss": 0.0648,
+      "step": 657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1206953114969196e-06,
+      "loss": 0.0581,
+      "step": 658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.093949129787165e-06,
+      "loss": 0.0378,
+      "step": 659
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.067297065473902e-06,
+      "loss": 0.0655,
+      "step": 660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0407394817781598e-06,
+      "loss": 0.0676,
+      "step": 661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.014276740633352e-06,
+      "loss": 0.0661,
+      "step": 662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9879092026803736e-06,
+      "loss": 0.064,
+      "step": 663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.961637227262655e-06,
+      "loss": 0.0527,
+      "step": 664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9354611724212768e-06,
+      "loss": 0.0646,
+      "step": 665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9093813948900886e-06,
+      "loss": 0.077,
+      "step": 666
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.883398250090861e-06,
+      "loss": 0.0674,
+      "step": 667
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8575120921284115e-06,
+      "loss": 0.0576,
+      "step": 668
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8317232737858034e-06,
+      "loss": 0.0618,
+      "step": 669
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8060321465195406e-06,
+      "loss": 0.0729,
+      "step": 670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.780439060454756e-06,
+      "loss": 0.0567,
+      "step": 671
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7549443643804585e-06,
+      "loss": 0.0569,
+      "step": 672
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7295484057447707e-06,
+      "loss": 0.0681,
+      "step": 673
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7042515306501992e-06,
+      "loss": 0.0607,
+      "step": 674
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6790540838489132e-06,
+      "loss": 0.0886,
+      "step": 675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6539564087380454e-06,
+      "loss": 0.0662,
+      "step": 676
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.628958847355029e-06,
+      "loss": 0.064,
+      "step": 677
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6040617403728985e-06,
+      "loss": 0.0744,
+      "step": 678
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.579265427095692e-06,
+      "loss": 0.0614,
+      "step": 679
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5545702454537945e-06,
+      "loss": 0.0764,
+      "step": 680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5299765319993464e-06,
+      "loss": 0.0643,
+      "step": 681
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5054846219016547e-06,
+      "loss": 0.0532,
+      "step": 682
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4810948489426213e-06,
+      "loss": 0.0597,
+      "step": 683
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4568075455122077e-06,
+      "loss": 0.0486,
+      "step": 684
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.43262304260388e-06,
+      "loss": 0.05,
+      "step": 685
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4085416698101207e-06,
+      "loss": 0.0668,
+      "step": 686
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3845637553179333e-06,
+      "loss": 0.0634,
+      "step": 687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3606896259043598e-06,
+      "loss": 0.0585,
+      "step": 688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3369196069320333e-06,
+      "loss": 0.0786,
+      "step": 689
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3132540223447465e-06,
+      "loss": 0.0639,
+      "step": 690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2896931946630307e-06,
+      "loss": 0.065,
+      "step": 691
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2662374449797663e-06,
+      "loss": 0.0598,
+      "step": 692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2428870929558012e-06,
+      "loss": 0.0755,
+      "step": 693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2196424568156073e-06,
+      "loss": 0.0684,
+      "step": 694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.196503853342915e-06,
+      "loss": 0.0604,
+      "step": 695
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1734715978764367e-06,
+      "loss": 0.0551,
+      "step": 696
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1505460043055316e-06,
+      "loss": 0.069,
+      "step": 697
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.127727385065951e-06,
+      "loss": 0.0563,
+      "step": 698
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.105016051135571e-06,
+      "loss": 0.058,
+      "step": 699
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0824123120301564e-06,
+      "loss": 0.0568,
+      "step": 700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.059916475799143e-06,
+      "loss": 0.0655,
+      "step": 701
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.037528849021441e-06,
+      "loss": 0.0738,
+      "step": 702
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.015249736801248e-06,
+      "loss": 0.0634,
+      "step": 703
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9930794427639134e-06,
+      "loss": 0.0565,
+      "step": 704
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9710182690517634e-06,
+      "loss": 0.0574,
+      "step": 705
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.949066516320025e-06,
+      "loss": 0.0668,
+      "step": 706
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9272244837326957e-06,
+      "loss": 0.0515,
+      "step": 707
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9054924689584864e-06,
+      "loss": 0.0726,
+      "step": 708
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8838707681667534e-06,
+      "loss": 0.0756,
+      "step": 709
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8623596760234664e-06,
+      "loss": 0.0536,
+      "step": 710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8409594856871971e-06,
+      "loss": 0.076,
+      "step": 711
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.819670488805111e-06,
+      "loss": 0.0781,
+      "step": 712
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7984929755090141e-06,
+      "loss": 0.0626,
+      "step": 713
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7774272344113775e-06,
+      "loss": 0.0608,
+      "step": 714
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7564735526014065e-06,
+      "loss": 0.0593,
+      "step": 715
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7356322156411487e-06,
+      "loss": 0.0584,
+      "step": 716
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7149035075615795e-06,
+      "loss": 0.0679,
+      "step": 717
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6942877108587397e-06,
+      "loss": 0.0768,
+      "step": 718
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.673785106489888e-06,
+      "loss": 0.0614,
+      "step": 719
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.653395973869668e-06,
+      "loss": 0.0618,
+      "step": 720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6331205908663006e-06,
+      "loss": 0.0586,
+      "step": 721
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6129592337977995e-06,
+      "loss": 0.0573,
+      "step": 722
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5929121774282087e-06,
+      "loss": 0.0572,
+      "step": 723
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5729796949638475e-06,
+      "loss": 0.0725,
+      "step": 724
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.553162058049591e-06,
+      "loss": 0.0569,
+      "step": 725
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5334595367651805e-06,
+      "loss": 0.0633,
+      "step": 726
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5138723996215233e-06,
+      "loss": 0.0763,
+      "step": 727
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.494400913557047e-06,
+      "loss": 0.0533,
+      "step": 728
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.475045343934054e-06,
+      "loss": 0.0561,
+      "step": 729
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4558059545351144e-06,
+      "loss": 0.0434,
+      "step": 730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4366830075594606e-06,
+      "loss": 0.0561,
+      "step": 731
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4176767636194122e-06,
+      "loss": 0.0702,
+      "step": 732
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3987874817368453e-06,
+      "loss": 0.0568,
+      "step": 733
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3800154193396365e-06,
+      "loss": 0.0754,
+      "step": 734
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3613608322581685e-06,
+      "loss": 0.0544,
+      "step": 735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3428239747218409e-06,
+      "loss": 0.0614,
+      "step": 736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3244050993556068e-06,
+      "loss": 0.0466,
+      "step": 737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3061044571765268e-06,
+      "loss": 0.0471,
+      "step": 738
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2879222975903494e-06,
+      "loss": 0.0666,
+      "step": 739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2698588683881185e-06,
+      "loss": 0.066,
+      "step": 740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2519144157427843e-06,
+      "loss": 0.076,
+      "step": 741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2340891842058511e-06,
+      "loss": 0.071,
+      "step": 742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.21638341670406e-06,
+      "loss": 0.0696,
+      "step": 743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1987973545360554e-06,
+      "loss": 0.0642,
+      "step": 744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1813312373691121e-06,
+      "loss": 0.0615,
+      "step": 745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1639853032358618e-06,
+      "loss": 0.0739,
+      "step": 746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.14675978853106e-06,
+      "loss": 0.0641,
+      "step": 747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1296549280083413e-06,
+      "loss": 0.0529,
+      "step": 748
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.112670954777043e-06,
+      "loss": 0.045,
+      "step": 749
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.095808100299025e-06,
+      "loss": 0.0688,
+      "step": 750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0790665943855028e-06,
+      "loss": 0.0682,
+      "step": 751
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0624466651939248e-06,
+      "loss": 0.0701,
+      "step": 752
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0459485392248625e-06,
+      "loss": 0.064,
+      "step": 753
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0295724413189212e-06,
+      "loss": 0.0546,
+      "step": 754
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0133185946536784e-06,
+      "loss": 0.0525,
+      "step": 755
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.971872207406397e-07,
+      "loss": 0.0548,
+      "step": 756
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.811785394222252e-07,
+      "loss": 0.0588,
+      "step": 757
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.652927688687607e-07,
+      "loss": 0.0665,
+      "step": 758
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.49530125575524e-07,
+      "loss": 0.0702,
+      "step": 759
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.338908243597766e-07,
+      "loss": 0.0475,
+      "step": 760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.183750783578438e-07,
+      "loss": 0.0885,
+      "step": 761
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.029830990222133e-07,
+      "loss": 0.0485,
+      "step": 762
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.87715096118642e-07,
+      "loss": 0.0517,
+      "step": 763
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.725712777233175e-07,
+      "loss": 0.0718,
+      "step": 764
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.575518502199953e-07,
+      "loss": 0.0683,
+      "step": 765
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.426570182972072e-07,
+      "loss": 0.0648,
+      "step": 766
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.278869849454718e-07,
+      "loss": 0.0725,
+      "step": 767
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.132419514545065e-07,
+      "loss": 0.0582,
+      "step": 768
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.98722117410512e-07,
+      "loss": 0.0683,
+      "step": 769
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843276806934353e-07,
+      "loss": 0.0621,
+      "step": 770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700588374742757e-07,
+      "loss": 0.0556,
+      "step": 771
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.55915782212413e-07,
+      "loss": 0.0775,
+      "step": 772
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.418987076529582e-07,
+      "loss": 0.0604,
+      "step": 773
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.280078048241279e-07,
+      "loss": 0.0714,
+      "step": 774
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.142432630346319e-07,
+      "loss": 0.0601,
+      "step": 775
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.006052698711075e-07,
+      "loss": 0.0534,
+      "step": 776
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.870940111955526e-07,
+      "loss": 0.0725,
+      "step": 777
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.737096711427915e-07,
+      "loss": 0.0484,
+      "step": 778
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.604524321179761e-07,
+      "loss": 0.0507,
+      "step": 779
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.47322474794091e-07,
+      "loss": 0.0471,
+      "step": 780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.343199781094933e-07,
+      "loss": 0.0557,
+      "step": 781
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.214451192654747e-07,
+      "loss": 0.0685,
+      "step": 782
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.086980737238458e-07,
+      "loss": 0.0656,
+      "step": 783
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.960790152045482e-07,
+      "loss": 0.057,
+      "step": 784
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.83588115683279e-07,
+      "loss": 0.0713,
+      "step": 785
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.71225545389158e-07,
+      "loss": 0.0599,
+      "step": 786
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.589914728024004e-07,
+      "loss": 0.06,
+      "step": 787
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.468860646520169e-07,
+      "loss": 0.0501,
+      "step": 788
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.34909485913554e-07,
+      "loss": 0.0506,
+      "step": 789
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.230618998068371e-07,
+      "loss": 0.0575,
+      "step": 790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.113434677937457e-07,
+      "loss": 0.0663,
+      "step": 791
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.997543495760126e-07,
+      "loss": 0.0605,
+      "step": 792
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.882947030930585e-07,
+      "loss": 0.0872,
+      "step": 793
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.769646845198217e-07,
+      "loss": 0.058,
+      "step": 794
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6576444826464173e-07,
+      "loss": 0.0695,
+      "step": 795
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5469414696715287e-07,
+      "loss": 0.0625,
+      "step": 796
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.437539314962047e-07,
+      "loss": 0.0751,
+      "step": 797
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.329439509477995e-07,
+      "loss": 0.0706,
+      "step": 798
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.222643526430703e-07,
+      "loss": 0.0691,
+      "step": 799
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.11715282126266e-07,
+      "loss": 0.0677,
+      "step": 800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.012968831627695e-07,
+      "loss": 0.0566,
+      "step": 801
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9100929773713937e-07,
+      "loss": 0.0496,
+      "step": 802
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.808526660511758e-07,
+      "loss": 0.0675,
+      "step": 803
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.708271265220087e-07,
+      "loss": 0.0459,
+      "step": 804
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.60932815780205e-07,
+      "loss": 0.0733,
+      "step": 805
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5116986866792104e-07,
+      "loss": 0.0714,
+      "step": 806
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4153841823705403e-07,
+      "loss": 0.0611,
+      "step": 807
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3203859574742816e-07,
+      "loss": 0.0762,
+      "step": 808
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.226705306650113e-07,
+      "loss": 0.0651,
+      "step": 809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1343435066015114e-07,
+      "loss": 0.0723,
+      "step": 810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.043301816058264e-07,
+      "loss": 0.0663,
+      "step": 811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.953581475759404e-07,
+      "loss": 0.0448,
+      "step": 812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.865183708436292e-07,
+      "loss": 0.0467,
+      "step": 813
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.778109718795907e-07,
+      "loss": 0.0731,
+      "step": 814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6923606935044477e-07,
+      "loss": 0.056,
+      "step": 815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.60793780117119e-07,
+      "loss": 0.0475,
+      "step": 816
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.524842192332522e-07,
+      "loss": 0.0569,
+      "step": 817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.443074999436257e-07,
+      "loss": 0.0584,
+      "step": 818
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3626373368262678e-07,
+      "loss": 0.0546,
+      "step": 819
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2835303007272324e-07,
+      "loss": 0.067,
+      "step": 820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.205754969229701e-07,
+      "loss": 0.046,
+      "step": 821
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1293124022754409e-07,
+      "loss": 0.0593,
+      "step": 822
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.05420364164296e-07,
+      "loss": 0.0646,
+      "step": 823
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9804297109333292e-07,
+      "loss": 0.0593,
+      "step": 824
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9079916155562038e-07,
+      "loss": 0.0605,
+      "step": 825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8368903427161354e-07,
+      "loss": 0.0574,
+      "step": 826
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.767126861399171e-07,
+      "loss": 0.0544,
+      "step": 827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6987021223595302e-07,
+      "loss": 0.0638,
+      "step": 828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.631617058106749e-07,
+      "loss": 0.067,
+      "step": 829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5658725828929688e-07,
+      "loss": 0.0532,
+      "step": 830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5014695927003885e-07,
+      "loss": 0.0583,
+      "step": 831
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4384089652291544e-07,
+      "loss": 0.0665,
+      "step": 832
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3766915598853347e-07,
+      "loss": 0.0574,
+      "step": 833
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3163182177692523e-07,
+      "loss": 0.064,
+      "step": 834
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2572897616639602e-07,
+      "loss": 0.0553,
+      "step": 835
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1996069960240943e-07,
+      "loss": 0.0548,
+      "step": 836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1432707069649051e-07,
+      "loss": 0.0627,
+      "step": 837
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0882816622514559e-07,
+      "loss": 0.0643,
+      "step": 838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0346406112882956e-07,
+      "loss": 0.0513,
+      "step": 839
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.823482851091359e-08,
+      "loss": 0.0397,
+      "step": 840
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.314053963669245e-08,
+      "loss": 0.0744,
+      "step": 841
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.818126393241644e-08,
+      "loss": 0.0502,
+      "step": 842
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.335706898433993e-08,
+      "loss": 0.0673,
+      "step": 843
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.86680205378043e-08,
+      "loss": 0.0656,
+      "step": 844
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.411418249634095e-08,
+      "loss": 0.0519,
+      "step": 845
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.96956169207963e-08,
+      "loss": 0.0627,
+      "step": 846
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.541238402849482e-08,
+      "loss": 0.0714,
+      "step": 847
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.126454219240852e-08,
+      "loss": 0.0714,
+      "step": 848
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7252147940369816e-08,
+      "loss": 0.0494,
+      "step": 849
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3375255954295494e-08,
+      "loss": 0.0562,
+      "step": 850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9633919069442815e-08,
+      "loss": 0.0654,
+      "step": 851
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.602818827369127e-08,
+      "loss": 0.0627,
+      "step": 852
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.25581127068464e-08,
+      "loss": 0.0708,
+      "step": 853
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9223739659970393e-08,
+      "loss": 0.059,
+      "step": 854
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.602511457473479e-08,
+      "loss": 0.0805,
+      "step": 855
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.296228104280874e-08,
+      "loss": 0.0836,
+      "step": 856
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0035280805255086e-08,
+      "loss": 0.0524,
+      "step": 857
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.724415375196965e-08,
+      "loss": 0.0728,
+      "step": 858
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4588937921132815e-08,
+      "loss": 0.0596,
+      "step": 859
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.206966949869327e-08,
+      "loss": 0.0628,
+      "step": 860
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.968638281787394e-08,
+      "loss": 0.0557,
+      "step": 861
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7439110358704603e-08,
+      "loss": 0.0541,
+      "step": 862
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5327882747578903e-08,
+      "loss": 0.0519,
+      "step": 863
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3352728756836908e-08,
+      "loss": 0.0596,
+      "step": 864
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1513675304373195e-08,
+      "loss": 0.0672,
+      "step": 865
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.810747453271596e-09,
+      "loss": 0.0487,
+      "step": 866
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.243968411461023e-09,
+      "loss": 0.0492,
+      "step": 867
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.813359531397945e-09,
+      "loss": 0.0586,
+      "step": 868
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.518940309779952e-09,
+      "loss": 0.0528,
+      "step": 869
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3607283872793e-09,
+      "loss": 0.0748,
+      "step": 870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3387395482975538e-09,
+      "loss": 0.0562,
+      "step": 871
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4529877207557507e-09,
+      "loss": 0.0531,
+      "step": 872
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7034849759023365e-09,
+      "loss": 0.0518,
+      "step": 873
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.09024152814885e-09,
+      "loss": 0.0573,
+      "step": 874
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.132657349322557e-10,
+      "loss": 0.0607,
+      "step": 875
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.725640965961507e-10,
+      "loss": 0.0422,
+      "step": 876
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.814125631082746e-11,
+      "loss": 0.0591,
+      "step": 877
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.0491,
+      "step": 878
+    },
+    {
+      "epoch": 2.0,
+      "step": 878,
+      "total_flos": 24109773520896.0,
+      "train_loss": 0.08651331942825491,
+      "train_runtime": 5421.1554,
+      "train_samples_per_second": 10.351,
+      "train_steps_per_second": 0.162
+    }
+  ],
+  "max_steps": 878,
+  "num_train_epochs": 2,
+  "total_flos": 24109773520896.0,
+  "trial_name": null,
+  "trial_params": null
+}