diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14317 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 2382,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7777777777777776e-07,
+      "loss": 1.3753,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 1.4095,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.1234,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 1.1039,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3888888888888892e-06,
+      "loss": 1.1757,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.1638,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.944444444444445e-06,
+      "loss": 1.0976,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.1601,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 1.0702,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 1.1146,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.055555555555556e-06,
+      "loss": 1.1073,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1339,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6111111111111115e-06,
+      "loss": 1.0755,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 1.0419,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.0515,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.9806,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.722222222222222e-06,
+      "loss": 1.0725,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 1.0675,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2777777777777785e-06,
+      "loss": 1.0665,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 1.0546,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.0393,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 1.0604,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3888888888888885e-06,
+      "loss": 1.0184,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0198,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.9763,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 1.0647,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.0404,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 0.9889,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.055555555555557e-06,
+      "loss": 1.08,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.0532,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.611111111111112e-06,
+      "loss": 1.0515,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.0821,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.1063,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 1.0613,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 1.0587,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 1.0521,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0277777777777777e-05,
+      "loss": 1.0106,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0555555555555557e-05,
+      "loss": 1.0364,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 1.0087,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.9911,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.138888888888889e-05,
+      "loss": 1.0746,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 1.0955,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1944444444444444e-05,
+      "loss": 0.9919,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 1.107,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.25e-05,
+      "loss": 1.0243,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 1.044,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3055555555555557e-05,
+      "loss": 1.0253,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.0447,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3611111111111113e-05,
+      "loss": 1.1005,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.9752,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 1.0652,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 1.0952,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4722222222222224e-05,
+      "loss": 0.9795,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.0751,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5277777777777777e-05,
+      "loss": 1.0217,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 1.0488,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 1.0799,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6111111111111115e-05,
+      "loss": 0.9951,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.638888888888889e-05,
+      "loss": 1.0324,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.0226,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6944444444444446e-05,
+      "loss": 1.0959,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 1.039,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.9827,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 1.1107,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8055555555555558e-05,
+      "loss": 1.0229,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 1.0655,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8611111111111114e-05,
+      "loss": 1.0291,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 1.0508,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 1.0822,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 1.0399,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9722222222222224e-05,
+      "loss": 1.0408,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.0705,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999075204468e-05,
+      "loss": 1.0717,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999963008195813e-05,
+      "loss": 0.9818,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999916768504724e-05,
+      "loss": 1.1086,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999985203305693e-05,
+      "loss": 1.0427,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999768801972174e-05,
+      "loss": 1.0553,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999667075404385e-05,
+      "loss": 1.0251,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999546853541728e-05,
+      "loss": 1.0666,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999408136606566e-05,
+      "loss": 1.0669,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999250924855458e-05,
+      "loss": 1.0973,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999075218579184e-05,
+      "loss": 1.0185,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998881018102735e-05,
+      "loss": 1.0915,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998668323785298e-05,
+      "loss": 1.0327,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998437136020274e-05,
+      "loss": 1.0682,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999818745523526e-05,
+      "loss": 1.0801,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997919281892066e-05,
+      "loss": 1.0023,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997632616486707e-05,
+      "loss": 1.04,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999732745954939e-05,
+      "loss": 1.0193,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997003811644534e-05,
+      "loss": 1.0408,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996661673370757e-05,
+      "loss": 1.1188,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996301045360874e-05,
+      "loss": 1.0742,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995921928281893e-05,
+      "loss": 1.0724,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995524322835035e-05,
+      "loss": 0.9667,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995108229755704e-05,
+      "loss": 1.0312,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99946736498135e-05,
+      "loss": 1.0651,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994220583812218e-05,
+      "loss": 1.0041,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999374903258985e-05,
+      "loss": 1.0412,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993258997018564e-05,
+      "loss": 1.0564,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999275047800474e-05,
+      "loss": 1.0484,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992223476488912e-05,
+      "loss": 1.0445,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991677993445832e-05,
+      "loss": 1.0562,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999111402988441e-05,
+      "loss": 1.0532,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990531586847756e-05,
+      "loss": 1.0229,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989930665413148e-05,
+      "loss": 1.0524,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989311266692046e-05,
+      "loss": 1.0759,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988673391830082e-05,
+      "loss": 1.1227,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988017042007066e-05,
+      "loss": 1.0867,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987342218436975e-05,
+      "loss": 0.9734,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986648922367957e-05,
+      "loss": 1.0744,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998593715508233e-05,
+      "loss": 1.1013,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985206917896563e-05,
+      "loss": 1.0691,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998445821216131e-05,
+      "loss": 1.0386,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983691039261358e-05,
+      "loss": 1.061,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982905400615672e-05,
+      "loss": 1.0199,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998210129767735e-05,
+      "loss": 0.9893,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998127873193367e-05,
+      "loss": 1.0822,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998043770490603e-05,
+      "loss": 1.0543,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979578218149992e-05,
+      "loss": 1.0518,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978700273255254e-05,
+      "loss": 1.0808,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997780387184565e-05,
+      "loss": 1.0303,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976889015579167e-05,
+      "loss": 1.0412,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975955706147902e-05,
+      "loss": 1.0516,
+      "step": 123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975003945278103e-05,
+      "loss": 1.0133,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997403373473014e-05,
+      "loss": 0.9543,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99730450762985e-05,
+      "loss": 1.0677,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972037971811802e-05,
+      "loss": 1.0258,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971012423132776e-05,
+      "loss": 1.0413,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969968432158266e-05,
+      "loss": 1.0642,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996890600081923e-05,
+      "loss": 1.0247,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996782513108073e-05,
+      "loss": 1.0785,
+      "step": 131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966725824941933e-05,
+      "loss": 1.1249,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996560808443611e-05,
+      "loss": 1.1129,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996447191163062e-05,
+      "loss": 1.0155,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963317308626916e-05,
+      "loss": 1.0085,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962144277560544e-05,
+      "loss": 1.0265,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960952820601135e-05,
+      "loss": 0.9788,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959742939952393e-05,
+      "loss": 1.0039,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958514637852103e-05,
+      "loss": 0.971,
+      "step": 139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957267916572122e-05,
+      "loss": 1.0537,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995600277841837e-05,
+      "loss": 1.0131,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954719225730847e-05,
+      "loss": 1.0482,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995341726088359e-05,
+      "loss": 1.0214,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995209688628471e-05,
+      "loss": 1.0193,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995075810437635e-05,
+      "loss": 1.0908,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949400917634715e-05,
+      "loss": 1.0918,
+      "step": 146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994802532857004e-05,
+      "loss": 0.9783,
+      "step": 147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946631339726614e-05,
+      "loss": 1.0278,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 1.0213,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943788173050743e-05,
+      "loss": 0.9924,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942339000477e-05,
+      "loss": 1.1012,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994087143864188e-05,
+      "loss": 1.0422,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939385490259773e-05,
+      "loss": 1.0369,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937881158079078e-05,
+      "loss": 1.0305,
+      "step": 154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9936358444882188e-05,
+      "loss": 1.0616,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99348173534855e-05,
+      "loss": 1.047,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933257886739416e-05,
+      "loss": 0.9724,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931680047528296e-05,
+      "loss": 1.0693,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930083838770504e-05,
+      "loss": 0.9897,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928469263418376e-05,
+      "loss": 1.0214,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992683632445821e-05,
+      "loss": 1.0,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992518502491028e-05,
+      "loss": 1.0463,
+      "step": 162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923515367828812e-05,
+      "loss": 1.0359,
+      "step": 163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992182735630199e-05,
+      "loss": 0.9741,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920120993451947e-05,
+      "loss": 1.04,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918396282434752e-05,
+      "loss": 0.9986,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991665322644042e-05,
+      "loss": 1.016,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991489182869289e-05,
+      "loss": 1.0238,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913112092450028e-05,
+      "loss": 0.9508,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911314021003614e-05,
+      "loss": 1.0507,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990949761767935e-05,
+      "loss": 1.0225,
+      "step": 171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907662885836836e-05,
+      "loss": 1.0627,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990580982886958e-05,
+      "loss": 1.0347,
+      "step": 173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903938450204972e-05,
+      "loss": 1.0141,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902048753304308e-05,
+      "loss": 1.0362,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900140741662745e-05,
+      "loss": 0.9904,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.9982,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896269788306967e-05,
+      "loss": 1.0061,
+      "step": 178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894306853752433e-05,
+      "loss": 1.018,
+      "step": 179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892325618776353e-05,
+      "loss": 0.9474,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98903260870432e-05,
+      "loss": 1.0989,
+      "step": 181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 1.0617,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988627214813277e-05,
+      "loss": 1.0634,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884217748453625e-05,
+      "loss": 1.0078,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882145067013652e-05,
+      "loss": 1.005,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880054107646467e-05,
+      "loss": 0.9824,
+      "step": 186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877944874219485e-05,
+      "loss": 1.0123,
+      "step": 187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875817370633923e-05,
+      "loss": 0.9919,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98736716008248e-05,
+      "loss": 0.9723,
+      "step": 189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987150756876091e-05,
+      "loss": 1.0069,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869325278444824e-05,
+      "loss": 1.0086,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986712473391289e-05,
+      "loss": 0.9846,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864905939235215e-05,
+      "loss": 0.987,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862668898515665e-05,
+      "loss": 1.0565,
+      "step": 194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986041361589184e-05,
+      "loss": 1.0129,
+      "step": 195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858140095535103e-05,
+      "loss": 1.0021,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985584834165053e-05,
+      "loss": 1.0107,
+      "step": 197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853538358476933e-05,
+      "loss": 1.041,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851210150286828e-05,
+      "loss": 1.0146,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848863721386456e-05,
+      "loss": 0.9785,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984649907611575e-05,
+      "loss": 1.0239,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844116218848335e-05,
+      "loss": 0.9492,
+      "step": 202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984171515399152e-05,
+      "loss": 0.9675,
+      "step": 203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98392958859863e-05,
+      "loss": 0.984,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836858419307325e-05,
+      "loss": 1.0201,
+      "step": 205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834402758462913e-05,
+      "loss": 1.0171,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831928907995032e-05,
+      "loss": 0.9976,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829436872479296e-05,
+      "loss": 1.0257,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982692665652495e-05,
+      "loss": 0.9712,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982439826477487e-05,
+      "loss": 1.0048,
+      "step": 210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982185170190554e-05,
+      "loss": 0.9814,
+      "step": 211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819286972627066e-05,
+      "loss": 0.992,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816704081683152e-05,
+      "loss": 0.9822,
+      "step": 213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814103033851078e-05,
+      "loss": 1.0232,
+      "step": 214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811483833941726e-05,
+      "loss": 0.9198,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980884648679955e-05,
+      "loss": 1.0002,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980619099730255e-05,
+      "loss": 1.0029,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803517370362308e-05,
+      "loss": 0.9706,
+      "step": 218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800825610923937e-05,
+      "loss": 1.0648,
+      "step": 219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798115723966087e-05,
+      "loss": 1.025,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795387714500946e-05,
+      "loss": 1.0053,
+      "step": 221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792641587574212e-05,
+      "loss": 0.9951,
+      "step": 222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97898773482651e-05,
+      "loss": 0.9304,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787095001686323e-05,
+      "loss": 0.957,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784294552984077e-05,
+      "loss": 1.0438,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.9693,
+      "step": 226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778639369961414e-05,
+      "loss": 0.9878,
+      "step": 227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775784646100768e-05,
+      "loss": 1.0243,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772911841036193e-05,
+      "loss": 0.9803,
+      "step": 229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97700209600812e-05,
+      "loss": 1.029,
+      "step": 230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767112008582736e-05,
+      "loss": 1.1119,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764184991921178e-05,
+      "loss": 1.0372,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761239915510302e-05,
+      "loss": 1.0,
+      "step": 233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97582767847973e-05,
+      "loss": 0.9891,
+      "step": 234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975529560526275e-05,
+      "loss": 1.0757,
+      "step": 235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975229638242062e-05,
+      "loss": 0.9528,
+      "step": 236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 1.0509,
+      "step": 237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746243829036308e-05,
+      "loss": 1.0063,
+      "step": 238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743190509688873e-05,
+      "loss": 1.1056,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740119169423337e-05,
+      "loss": 0.9329,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737029813920414e-05,
+      "loss": 1.0393,
+      "step": 241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973392244889415e-05,
+      "loss": 0.9744,
+      "step": 242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730797080091904e-05,
+      "loss": 0.9752,
+      "step": 243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727653713294325e-05,
+      "loss": 1.0531,
+      "step": 244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972449235431536e-05,
+      "loss": 0.9769,
+      "step": 245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721313009002228e-05,
+      "loss": 1.0232,
+      "step": 246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718115683235418e-05,
+      "loss": 1.0316,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714900382928674e-05,
+      "loss": 0.9918,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971166711402899e-05,
+      "loss": 1.0564,
+      "step": 249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970841588251659e-05,
+      "loss": 1.0005,
+      "step": 250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705146694404923e-05,
+      "loss": 1.1,
+      "step": 251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701859555740647e-05,
+      "loss": 1.0852,
+      "step": 252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698554472603628e-05,
+      "loss": 0.9677,
+      "step": 253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695231451106914e-05,
+      "loss": 0.8954,
+      "step": 254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969189049739674e-05,
+      "loss": 1.0191,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688531617652503e-05,
+      "loss": 1.0084,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968515481808676e-05,
+      "loss": 0.9986,
+      "step": 257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681760104945203e-05,
+      "loss": 0.96,
+      "step": 258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678347484506667e-05,
+      "loss": 0.9662,
+      "step": 259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674916963083103e-05,
+      "loss": 0.9937,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671468547019575e-05,
+      "loss": 1.0179,
+      "step": 261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966800224269424e-05,
+      "loss": 0.9517,
+      "step": 262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966451805651834e-05,
+      "loss": 0.9778,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661015994936204e-05,
+      "loss": 1.0035,
+      "step": 264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657496064425206e-05,
+      "loss": 1.0059,
+      "step": 265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9653958271495785e-05,
+      "loss": 0.9607,
+      "step": 266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96504026226914e-05,
+      "loss": 0.9569,
+      "step": 267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964682912458856e-05,
+      "loss": 1.0273,
+      "step": 268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643237783796767e-05,
+      "loss": 0.9888,
+      "step": 269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639628606958535e-05,
+      "loss": 0.9603,
+      "step": 270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636001600749365e-05,
+      "loss": 1.0347,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632356771877735e-05,
+      "loss": 0.9789,
+      "step": 272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628694127085094e-05,
+      "loss": 0.9799,
+      "step": 273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9625013673145826e-05,
+      "loss": 0.9912,
+      "step": 274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621315416867274e-05,
+      "loss": 0.9926,
+      "step": 275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617599365089693e-05,
+      "loss": 1.0224,
+      "step": 276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961386552468627e-05,
+      "loss": 0.9174,
+      "step": 277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961011390256307e-05,
+      "loss": 1.0108,
+      "step": 278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606344505659072e-05,
+      "loss": 0.9758,
+      "step": 279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960255734094611e-05,
+      "loss": 1.0371,
+      "step": 280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598752415428893e-05,
+      "loss": 0.9727,
+      "step": 281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 1.0446,
+      "step": 282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591089310164752e-05,
+      "loss": 0.9592,
+      "step": 283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587231144591444e-05,
+      "loss": 0.9786,
+      "step": 284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583355246561076e-05,
+      "loss": 0.9837,
+      "step": 285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579461623242473e-05,
+      "loss": 0.9976,
+      "step": 286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957555028183725e-05,
+      "loss": 1.0158,
+      "step": 287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957162122957978e-05,
+      "loss": 1.0175,
+      "step": 288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956767447373722e-05,
+      "loss": 0.9767,
+      "step": 289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9563710021609438e-05,
+      "loss": 1.0287,
+      "step": 290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559727880529057e-05,
+      "loss": 0.9163,
+      "step": 291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.9583,
+      "step": 292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955171056100453e-05,
+      "loss": 1.0176,
+      "step": 293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547675397389144e-05,
+      "loss": 0.9932,
+      "step": 294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543622574478657e-05,
+      "loss": 0.9823,
+      "step": 295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539552099769128e-05,
+      "loss": 0.99,
+      "step": 296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535463980789277e-05,
+      "loss": 0.9803,
+      "step": 297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953135822510045e-05,
+      "loss": 1.0359,
+      "step": 298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952723484029662e-05,
+      "loss": 0.9371,
+      "step": 299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523093834004358e-05,
+      "loss": 0.9813,
+      "step": 300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951893521388283e-05,
+      "loss": 0.9489,
+      "step": 301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514758987623784e-05,
+      "loss": 0.9289,
+      "step": 302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.9375,
+      "step": 303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506353747622943e-05,
+      "loss": 1.0032,
+      "step": 304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502124749427403e-05,
+      "loss": 1.0232,
+      "step": 305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949787817618683e-05,
+      "loss": 0.993,
+      "step": 306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493614035755647e-05,
+      "loss": 1.063,
+      "step": 307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489332336020776e-05,
+      "loss": 1.0367,
+      "step": 308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485033084901606e-05,
+      "loss": 0.946,
+      "step": 309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480716290349998e-05,
+      "loss": 1.0679,
+      "step": 310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947638196035025e-05,
+      "loss": 1.0827,
+      "step": 311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472030102919102e-05,
+      "loss": 1.0027,
+      "step": 312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467660726105713e-05,
+      "loss": 0.9434,
+      "step": 313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9463273837991643e-05,
+      "loss": 0.969,
+      "step": 314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945886944669084e-05,
+      "loss": 0.9593,
+      "step": 315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454447560349628e-05,
+      "loss": 1.0031,
+      "step": 316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450008187146685e-05,
+      "loss": 1.0413,
+      "step": 317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944555133529304e-05,
+      "loss": 0.9587,
+      "step": 318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944107701303205e-05,
+      "loss": 0.9996,
+      "step": 319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436585228639368e-05,
+      "loss": 1.0608,
+      "step": 320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943207599042297e-05,
+      "loss": 0.9899,
+      "step": 321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94275493067231e-05,
+      "loss": 0.977,
+      "step": 322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942300518591227e-05,
+      "loss": 1.006,
+      "step": 323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941844363639525e-05,
+      "loss": 0.982,
+      "step": 324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413864666609036e-05,
+      "loss": 1.0071,
+      "step": 325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940926828502285e-05,
+      "loss": 0.993,
+      "step": 326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404654500138117e-05,
+      "loss": 0.9914,
+      "step": 327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400023320488458e-05,
+      "loss": 1.008,
+      "step": 328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395374754639657e-05,
+      "loss": 1.0271,
+      "step": 329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939070881118966e-05,
+      "loss": 1.0351,
+      "step": 330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938602549876856e-05,
+      "loss": 1.0,
+      "step": 331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938132482603856e-05,
+      "loss": 1.0092,
+      "step": 332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937660680169399e-05,
+      "loss": 1.0312,
+      "step": 333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371871434461265e-05,
+      "loss": 0.9642,
+      "step": 334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367118733098876e-05,
+      "loss": 1.0035,
+      "step": 335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362348706397374e-05,
+      "loss": 1.0355,
+      "step": 336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935756136317936e-05,
+      "loss": 1.0127,
+      "step": 337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352756712299467e-05,
+      "loss": 1.0631,
+      "step": 338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347934762644326e-05,
+      "loss": 0.9229,
+      "step": 339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343095523132577e-05,
+      "loss": 0.9888,
+      "step": 340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338239002714833e-05,
+      "loss": 0.9635,
+      "step": 341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333365210373668e-05,
+      "loss": 1.0081,
+      "step": 342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932847415512361e-05,
+      "loss": 1.0725,
+      "step": 343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323565846011107e-05,
+      "loss": 1.0143,
+      "step": 344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318640292114526e-05,
+      "loss": 0.9558,
+      "step": 345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313697502544125e-05,
+      "loss": 0.9323,
+      "step": 346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308737486442045e-05,
+      "loss": 0.9584,
+      "step": 347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303760252982287e-05,
+      "loss": 0.9939,
+      "step": 348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92987658113707e-05,
+      "loss": 1.0367,
+      "step": 349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293754170844955e-05,
+      "loss": 1.0437,
+      "step": 350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288725340674536e-05,
+      "loss": 0.9287,
+      "step": 351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283679330160726e-05,
+      "loss": 1.011,
+      "step": 352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278616148636585e-05,
+      "loss": 0.95,
+      "step": 353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927353580546692e-05,
+      "loss": 0.9897,
+      "step": 354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268438310048293e-05,
+      "loss": 0.9903,
+      "step": 355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926332367180898e-05,
+      "loss": 0.9597,
+      "step": 356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925819190020898e-05,
+      "loss": 1.0048,
+      "step": 357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253043004739967e-05,
+      "loss": 1.0251,
+      "step": 358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247876994925293e-05,
+      "loss": 0.9652,
+      "step": 359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242693880319962e-05,
+      "loss": 1.0469,
+      "step": 360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237493670510618e-05,
+      "loss": 0.9144,
+      "step": 361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232276375115517e-05,
+      "loss": 1.0232,
+      "step": 362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922704200378453e-05,
+      "loss": 0.8986,
+      "step": 363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221790566199098e-05,
+      "loss": 1.0124,
+      "step": 364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216522072072236e-05,
+      "loss": 0.9552,
+      "step": 365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92112365311485e-05,
+      "loss": 1.0393,
+      "step": 366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920593395320399e-05,
+      "loss": 0.9683,
+      "step": 367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920061434804629e-05,
+      "loss": 0.9619,
+      "step": 368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919527772551451e-05,
+      "loss": 0.9795,
+      "step": 369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189924095479207e-05,
+      "loss": 1.033,
+      "step": 370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184553467842413e-05,
+      "loss": 0.9468,
+      "step": 371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179165852537596e-05,
+      "loss": 1.0507,
+      "step": 372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173761259529634e-05,
+      "loss": 0.9865,
+      "step": 373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916833969881482e-05,
+      "loss": 0.981,
+      "step": 374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916290118042082e-05,
+      "loss": 0.9299,
+      "step": 375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157445714406673e-05,
+      "loss": 0.9647,
+      "step": 376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151973310862754e-05,
+      "loss": 0.9746,
+      "step": 377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914648397991078e-05,
+      "loss": 1.0559,
+      "step": 378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140977731703762e-05,
+      "loss": 0.9615,
+      "step": 379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 1.0033,
+      "step": 380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129914524293102e-05,
+      "loss": 0.9955,
+      "step": 381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124357585551872e-05,
+      "loss": 1.0482,
+      "step": 382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118783770480377e-05,
+      "loss": 0.9858,
+      "step": 383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91131930893879e-05,
+      "loss": 1.0108,
+      "step": 384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910758555261492e-05,
+      "loss": 0.9476,
+      "step": 385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101961170533077e-05,
+      "loss": 0.9559,
+      "step": 386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.9967,
+      "step": 387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090661912085186e-05,
+      "loss": 0.9501,
+      "step": 388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084987056618144e-05,
+      "loss": 0.9697,
+      "step": 389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079295397640215e-05,
+      "loss": 1.042,
+      "step": 390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907358694567865e-05,
+      "loss": 0.9554,
+      "step": 391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067861711291744e-05,
+      "loss": 0.9701,
+      "step": 392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062119705068843e-05,
+      "loss": 1.0239,
+      "step": 393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905636093763031e-05,
+      "loss": 0.9964,
+      "step": 394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050585419627507e-05,
+      "loss": 0.9414,
+      "step": 395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044793161742783e-05,
+      "loss": 0.9458,
+      "step": 396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038984174689447e-05,
+      "loss": 0.9895,
+      "step": 397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033158469211747e-05,
+      "loss": 0.9795,
+      "step": 398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902731605608486e-05,
+      "loss": 0.9401,
+      "step": 399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021456946114854e-05,
+      "loss": 0.9402,
+      "step": 400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9015581150138693e-05,
+      "loss": 0.9585,
+      "step": 401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 1.0112,
+      "step": 402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003779543670017e-05,
+      "loss": 1.0436,
+      "step": 403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899785375500565e-05,
+      "loss": 0.9932,
+      "step": 404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899191132399138e-05,
+      "loss": 0.907,
+      "step": 405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898595226161827e-05,
+      "loss": 0.9548,
+      "step": 406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979976578908146e-05,
+      "loss": 0.968,
+      "step": 407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973984286913584e-05,
+      "loss": 0.9924,
+      "step": 408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8967975396717873e-05,
+      "loss": 0.9911,
+      "step": 409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961949919435e-05,
+      "loss": 0.9938,
+      "step": 410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895590786620963e-05,
+      "loss": 0.9691,
+      "step": 411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89498492482171e-05,
+      "loss": 1.036,
+      "step": 412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943774076663372e-05,
+      "loss": 0.9633,
+      "step": 413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937682362785025e-05,
+      "loss": 0.966,
+      "step": 414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893157411784924e-05,
+      "loss": 0.9701,
+      "step": 415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925449353153776e-05,
+      "loss": 0.9481,
+      "step": 416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891930808002694e-05,
+      "loss": 0.9798,
+      "step": 417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913150309827576e-05,
+      "loss": 0.9846,
+      "step": 418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906976053945043e-05,
+      "loss": 1.011,
+      "step": 419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890078532379919e-05,
+      "loss": 0.977,
+      "step": 420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889457813084033e-05,
+      "loss": 0.8985,
+      "step": 421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888354486549238e-05,
+      "loss": 0.9475,
+      "step": 422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882114402437105e-05,
+      "loss": 1.0425,
+      "step": 423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875857890045544e-05,
+      "loss": 0.9531,
+      "step": 424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8869584960946535e-05,
+      "loss": 1.0318,
+      "step": 425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863295626742438e-05,
+      "loss": 0.9958,
+      "step": 426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856989899065943e-05,
+      "loss": 1.0072,
+      "step": 427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850667789580073e-05,
+      "loss": 1.0116,
+      "step": 428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844329309978146e-05,
+      "loss": 0.9818,
+      "step": 429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837974471983754e-05,
+      "loss": 1.0041,
+      "step": 430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831603287350746e-05,
+      "loss": 1.016,
+      "step": 431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825215767863215e-05,
+      "loss": 0.9299,
+      "step": 432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8818811925335455e-05,
+      "loss": 1.0113,
+      "step": 433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812391771611956e-05,
+      "loss": 0.9543,
+      "step": 434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880595531856738e-05,
+      "loss": 0.9597,
+      "step": 435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799502578106533e-05,
+      "loss": 0.9883,
+      "step": 436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8793033562164344e-05,
+      "loss": 1.0129,
+      "step": 437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786548282705847e-05,
+      "loss": 1.0075,
+      "step": 438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780046751726154e-05,
+      "loss": 0.9693,
+      "step": 439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877352898125045e-05,
+      "loss": 0.9477,
+      "step": 440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876699498333393e-05,
+      "loss": 0.9282,
+      "step": 441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876044477006183e-05,
+      "loss": 0.8773,
+      "step": 442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753878353549357e-05,
+      "loss": 0.9544,
+      "step": 443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747295745941705e-05,
+      "loss": 1.0243,
+      "step": 444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8740696959413996e-05,
+      "loss": 1.024,
+      "step": 445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87340820061713e-05,
+      "loss": 0.9891,
+      "step": 446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727450898448562e-05,
+      "loss": 1.0099,
+      "step": 447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720803648510627e-05,
+      "loss": 0.9589,
+      "step": 448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8714140268652187e-05,
+      "loss": 0.9731,
+      "step": 449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707460771197773e-05,
+      "loss": 0.9447,
+      "step": 450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870076516850172e-05,
+      "loss": 0.9931,
+      "step": 451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8694053472948154e-05,
+      "loss": 1.0154,
+      "step": 452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687325696950972e-05,
+      "loss": 1.0124,
+      "step": 453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680581852953806e-05,
+      "loss": 1.0111,
+      "step": 454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673821953430007e-05,
+      "loss": 1.002,
+      "step": 455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8667046010882627e-05,
+      "loss": 1.0348,
+      "step": 456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 1.0363,
+      "step": 457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653446046877665e-05,
+      "loss": 0.9211,
+      "step": 458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646622050574456e-05,
+      "loss": 0.9316,
+      "step": 459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8639782061556362e-05,
+      "loss": 0.9497,
+      "step": 460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8632926092474565e-05,
+      "loss": 0.8993,
+      "step": 461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626054156009807e-05,
+      "loss": 0.9596,
+      "step": 462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8619166264872363e-05,
+      "loss": 0.9771,
+      "step": 463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861226243180201e-05,
+      "loss": 1.0138,
+      "step": 464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605342669568013e-05,
+      "loss": 0.9653,
+      "step": 465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598406990969108e-05,
+      "loss": 0.9534,
+      "step": 466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591455408833462e-05,
+      "loss": 1.0245,
+      "step": 467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584487936018663e-05,
+      "loss": 1.0788,
+      "step": 468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857750458541168e-05,
+      "loss": 0.9782,
+      "step": 469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857050536992886e-05,
+      "loss": 0.9864,
+      "step": 470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856349030251589e-05,
+      "loss": 0.9667,
+      "step": 471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556459396147777e-05,
+      "loss": 0.9213,
+      "step": 472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549412663828813e-05,
+      "loss": 0.9498,
+      "step": 473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542350118592585e-05,
+      "loss": 0.9528,
+      "step": 474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8535271773501905e-05,
+      "loss": 0.9668,
+      "step": 475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528177641648817e-05,
+      "loss": 0.9525,
+      "step": 476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521067736154567e-05,
+      "loss": 0.9512,
+      "step": 477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513942070169572e-05,
+      "loss": 1.0068,
+      "step": 478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506800656873397e-05,
+      "loss": 0.9769,
+      "step": 479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499643509474738e-05,
+      "loss": 0.9766,
+      "step": 480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8492470641211396e-05,
+      "loss": 1.0235,
+      "step": 481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485282065350237e-05,
+      "loss": 0.8887,
+      "step": 482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478077795187188e-05,
+      "loss": 1.0392,
+      "step": 483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847085784404721e-05,
+      "loss": 1.0398,
+      "step": 484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8463622225284242e-05,
+      "loss": 1.004,
+      "step": 485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456370952281243e-05,
+      "loss": 0.9614,
+      "step": 486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844910403845009e-05,
+      "loss": 0.9358,
+      "step": 487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8441821497231597e-05,
+      "loss": 0.9644,
+      "step": 488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84345233420955e-05,
+      "loss": 0.9807,
+      "step": 489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427209586540392e-05,
+      "loss": 0.9552,
+      "step": 490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419880244093735e-05,
+      "loss": 0.9427,
+      "step": 491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 1.0445,
+      "step": 492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405174852779717e-05,
+      "loss": 1.0056,
+      "step": 493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397798831111316e-05,
+      "loss": 1.0009,
+      "step": 494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390407276949235e-05,
+      "loss": 0.9659,
+      "step": 495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383000203964826e-05,
+      "loss": 0.9422,
+      "step": 496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375577625858148e-05,
+      "loss": 1.0116,
+      "step": 497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836813955635793e-05,
+      "loss": 0.9563,
+      "step": 498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836068600922156e-05,
+      "loss": 0.9156,
+      "step": 499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353216998235057e-05,
+      "loss": 0.9198,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834573253721303e-05,
+      "loss": 0.9774,
+      "step": 501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338232639998672e-05,
+      "loss": 0.9469,
+      "step": 502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8330717320463732e-05,
+      "loss": 0.9689,
+      "step": 503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8323186592508474e-05,
+      "loss": 0.9422,
+      "step": 504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315640470061667e-05,
+      "loss": 1.0335,
+      "step": 505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308078967080547e-05,
+      "loss": 0.9574,
+      "step": 506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300502097550807e-05,
+      "loss": 0.9709,
+      "step": 507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8292909875486556e-05,
+      "loss": 0.9908,
+      "step": 508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82853023149303e-05,
+      "loss": 0.9798,
+      "step": 509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277679429952913e-05,
+      "loss": 0.9858,
+      "step": 510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270041234653617e-05,
+      "loss": 0.9446,
+      "step": 511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.9501,
+      "step": 512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825471896962774e-05,
+      "loss": 0.9733,
+      "step": 513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8247034928241086e-05,
+      "loss": 1.0508,
+      "step": 514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8239335633212316e-05,
+      "loss": 0.9289,
+      "step": 515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231621098781983e-05,
+      "loss": 0.9225,
+      "step": 516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223891339218815e-05,
+      "loss": 0.897,
+      "step": 517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216146368819713e-05,
+      "loss": 0.9616,
+      "step": 518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8208386201909702e-05,
+      "loss": 0.9671,
+      "step": 519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200610852841913e-05,
+      "loss": 0.9585,
+      "step": 520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192820335997568e-05,
+      "loss": 0.9082,
+      "step": 521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8185014665785936e-05,
+      "loss": 0.9371,
+      "step": 522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177193856644315e-05,
+      "loss": 1.0159,
+      "step": 523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8169357923038e-05,
+      "loss": 0.9892,
+      "step": 524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161506879460274e-05,
+      "loss": 1.0045,
+      "step": 525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153640740432344e-05,
+      "loss": 1.0151,
+      "step": 526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814575952050336e-05,
+      "loss": 0.9909,
+      "step": 527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137863234250346e-05,
+      "loss": 0.9681,
+      "step": 528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8129951896278216e-05,
+      "loss": 0.9473,
+      "step": 529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81220255212197e-05,
+      "loss": 0.938,
+      "step": 530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114084123735355e-05,
+      "loss": 0.9624,
+      "step": 531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810612771851352e-05,
+      "loss": 1.0112,
+      "step": 532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809815632027028e-05,
+      "loss": 0.9893,
+      "step": 533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.9184,
+      "step": 534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8082168603722632e-05,
+      "loss": 0.9237,
+      "step": 535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8074152314988957e-05,
+      "loss": 0.9121,
+      "step": 536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8066121092375303e-05,
+      "loss": 0.9221,
+      "step": 537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805807495073615e-05,
+      "loss": 0.9244,
+      "step": 538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050013904953572e-05,
+      "loss": 1.038,
+      "step": 539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8041937969937206e-05,
+      "loss": 0.9071,
+      "step": 540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803384716062423e-05,
+      "loss": 0.9808,
+      "step": 541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025741491979326e-05,
+      "loss": 0.9741,
+      "step": 542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017620978994676e-05,
+      "loss": 0.8941,
+      "step": 543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8009485636689903e-05,
+      "loss": 0.8897,
+      "step": 544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001335480112067e-05,
+      "loss": 0.9123,
+      "step": 545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7993170524335614e-05,
+      "loss": 0.9699,
+      "step": 546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798499078446239e-05,
+      "loss": 0.9119,
+      "step": 547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7976796275621556e-05,
+      "loss": 1.0456,
+      "step": 548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7968587012969606e-05,
+      "loss": 0.9886,
+      "step": 549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796036301169032e-05,
+      "loss": 1.0272,
+      "step": 550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952124286994735e-05,
+      "loss": 1.0079,
+      "step": 551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7943870854121126e-05,
+      "loss": 0.9688,
+      "step": 552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935602728334968e-05,
+      "loss": 1.018,
+      "step": 553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927319924928908e-05,
+      "loss": 0.9513,
+      "step": 554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919022459222754e-05,
+      "loss": 1.054,
+      "step": 555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910710346563417e-05,
+      "loss": 0.9938,
+      "step": 556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7902383602324905e-05,
+      "loss": 1.0158,
+      "step": 557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894042241908293e-05,
+      "loss": 0.9314,
+      "step": 558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788568628074169e-05,
+      "loss": 0.9169,
+      "step": 559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7877315734280204e-05,
+      "loss": 1.0225,
+      "step": 560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786893061800592e-05,
+      "loss": 0.9723,
+      "step": 561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860530947427878e-05,
+      "loss": 0.9372,
+      "step": 562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7852116738082032e-05,
+      "loss": 1.0268,
+      "step": 563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7843688005531227e-05,
+      "loss": 0.9625,
+      "step": 564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7835244765365177e-05,
+      "loss": 0.9757,
+      "step": 565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782678703320042e-05,
+      "loss": 0.9728,
+      "step": 566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 1.0215,
+      "step": 567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780982815547494e-05,
+      "loss": 1.0294,
+      "step": 568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780132704128121e-05,
+      "loss": 0.973,
+      "step": 569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779281149782269e-05,
+      "loss": 0.9302,
+      "step": 570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784281540849656e-05,
+      "loss": 0.9521,
+      "step": 571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777573718613904e-05,
+      "loss": 0.9514,
+      "step": 572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7767178449494403e-05,
+      "loss": 0.9691,
+      "step": 573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7758605346745907e-05,
+      "loss": 0.9392,
+      "step": 574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775001789375029e-05,
+      "loss": 1.0039,
+      "step": 575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7741416106390828e-05,
+      "loss": 1.0269,
+      "step": 576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732800000577304e-05,
+      "loss": 1.0255,
+      "step": 577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724169592245996e-05,
+      "loss": 0.9378,
+      "step": 578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771552489735963e-05,
+      "loss": 0.9787,
+      "step": 579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7706865931907352e-05,
+      "loss": 0.9235,
+      "step": 580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698192711904715e-05,
+      "loss": 0.9984,
+      "step": 581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768950525339362e-05,
+      "loss": 1.0416,
+      "step": 582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768080357244232e-05,
+      "loss": 0.9586,
+      "step": 583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7672087685145362e-05,
+      "loss": 0.8792,
+      "step": 584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663357607623577e-05,
+      "loss": 0.9431,
+      "step": 585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7654613356024032e-05,
+      "loss": 0.9765,
+      "step": 586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7645854946520024e-05,
+      "loss": 0.993,
+      "step": 587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7637082395311024e-05,
+      "loss": 0.9783,
+      "step": 588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7628295718622666e-05,
+      "loss": 1.0339,
+      "step": 589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761949493270671e-05,
+      "loss": 0.9383,
+      "step": 590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610680053841008e-05,
+      "loss": 0.9945,
+      "step": 591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7601851098329484e-05,
+      "loss": 0.9832,
+      "step": 592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759300808250209e-05,
+      "loss": 0.9317,
+      "step": 593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75841510227148e-05,
+      "loss": 0.8717,
+      "step": 594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757527993534954e-05,
+      "loss": 0.9547,
+      "step": 595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75663948368142e-05,
+      "loss": 1.0007,
+      "step": 596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.9598,
+      "step": 597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548582671994365e-05,
+      "loss": 0.9864,
+      "step": 598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7539655638655088e-05,
+      "loss": 0.9519,
+      "step": 599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530714660036112e-05,
+      "loss": 0.8844,
+      "step": 600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521759752674588e-05,
+      "loss": 0.8844,
+      "step": 601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7512790933133435e-05,
+      "loss": 0.928,
+      "step": 602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503808218001305e-05,
+      "loss": 0.9162,
+      "step": 603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494811623892543e-05,
+      "loss": 0.958,
+      "step": 604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485801167447166e-05,
+      "loss": 0.9967,
+      "step": 605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476776865330847e-05,
+      "loss": 0.9027,
+      "step": 606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746773873423484e-05,
+      "loss": 0.954,
+      "step": 607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7458686790876e-05,
+      "loss": 0.9691,
+      "step": 608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7449621051996714e-05,
+      "loss": 0.9665,
+      "step": 609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7440541534364896e-05,
+      "loss": 0.9902,
+      "step": 610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 0.9317,
+      "step": 611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74223412300427e-05,
+      "loss": 0.9872,
+      "step": 612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741322047701544e-05,
+      "loss": 0.9489,
+      "step": 613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7404086012561824e-05,
+      "loss": 0.9604,
+      "step": 614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739493785357688e-05,
+      "loss": 0.9511,
+      "step": 615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385776016980957e-05,
+      "loss": 1.0024,
+      "step": 616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737660051971971e-05,
+      "loss": 0.9499,
+      "step": 617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736741137876405e-05,
+      "loss": 0.9778,
+      "step": 618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7358208611110136e-05,
+      "loss": 0.9372,
+      "step": 619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7348992233779326e-05,
+      "loss": 0.9715,
+      "step": 620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7339762263818144e-05,
+      "loss": 0.9496,
+      "step": 621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.9185,
+      "step": 622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7321261614316465e-05,
+      "loss": 1.0423,
+      "step": 623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311990968994598e-05,
+      "loss": 0.9978,
+      "step": 624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302706799479575e-05,
+      "loss": 1.0011,
+      "step": 625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293409122943306e-05,
+      "loss": 0.8799,
+      "step": 626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284097956582694e-05,
+      "loss": 0.939,
+      "step": 627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7274773317619583e-05,
+      "loss": 1.0079,
+      "step": 628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7265435223300753e-05,
+      "loss": 0.9944,
+      "step": 629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725608369089785e-05,
+      "loss": 0.9111,
+      "step": 630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724671873770739e-05,
+      "loss": 0.8787,
+      "step": 631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72373403810507e-05,
+      "loss": 1.0348,
+      "step": 632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7227948638273918e-05,
+      "loss": 0.9751,
+      "step": 633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721854352674791e-05,
+      "loss": 0.9396,
+      "step": 634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209125063868303e-05,
+      "loss": 0.9486,
+      "step": 635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199693267055392e-05,
+      "loss": 0.987,
+      "step": 636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7190248153754147e-05,
+      "loss": 0.9094,
+      "step": 637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7180789741434163e-05,
+      "loss": 0.9248,
+      "step": 638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7171318047589637e-05,
+      "loss": 0.956,
+      "step": 639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7161833089739328e-05,
+      "loss": 0.9257,
+      "step": 640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152334885426533e-05,
+      "loss": 0.9114,
+      "step": 641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7142823452219036e-05,
+      "loss": 0.987,
+      "step": 642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7133298807709113e-05,
+      "loss": 1.0171,
+      "step": 643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712376096951345e-05,
+      "loss": 1.0031,
+      "step": 644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114209955273154e-05,
+      "loss": 1.0003,
+      "step": 645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710464578265369e-05,
+      "loss": 0.9851,
+      "step": 646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709506846934487e-05,
+      "loss": 0.9683,
+      "step": 647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085478033060808e-05,
+      "loss": 1.0059,
+      "step": 648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075874491539886e-05,
+      "loss": 0.9977,
+      "step": 649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706625786254473e-05,
+      "loss": 0.9265,
+      "step": 650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705662816386217e-05,
+      "loss": 0.9143,
+      "step": 651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7046985413303215e-05,
+      "loss": 0.9566,
+      "step": 652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7037329628703005e-05,
+      "loss": 0.9103,
+      "step": 653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7027660827920798e-05,
+      "loss": 1.0034,
+      "step": 654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017979028839918e-05,
+      "loss": 0.9302,
+      "step": 655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700828424936773e-05,
+      "loss": 0.9899,
+      "step": 656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699857650743562e-05,
+      "loss": 0.9822,
+      "step": 657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6988855820998935e-05,
+      "loss": 1.0186,
+      "step": 658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979122208036972e-05,
+      "loss": 0.9385,
+      "step": 659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696937568655294e-05,
+      "loss": 0.9022,
+      "step": 660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6959616274573902e-05,
+      "loss": 0.9937,
+      "step": 661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6949843990150798e-05,
+      "loss": 0.9748,
+      "step": 662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6940058851358344e-05,
+      "loss": 0.9319,
+      "step": 663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6930260876295052e-05,
+      "loss": 0.9513,
+      "step": 664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692045008308317e-05,
+      "loss": 0.9559,
+      "step": 665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691062648986865e-05,
+      "loss": 0.9283,
+      "step": 666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900790114821122e-05,
+      "loss": 0.9971,
+      "step": 667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890940976133857e-05,
+      "loss": 0.939,
+      "step": 668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6881079092023742e-05,
+      "loss": 0.961,
+      "step": 669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687120448073122e-05,
+      "loss": 0.9294,
+      "step": 670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6861317160520286e-05,
+      "loss": 0.9221,
+      "step": 671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851417149678442e-05,
+      "loss": 0.9495,
+      "step": 672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684150446651666e-05,
+      "loss": 1.0673,
+      "step": 673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6831579129369347e-05,
+      "loss": 0.9481,
+      "step": 674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682164115659432e-05,
+      "loss": 0.9418,
+      "step": 675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681169056657276e-05,
+      "loss": 1.0706,
+      "step": 676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6801727377709195e-05,
+      "loss": 1.0206,
+      "step": 677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679175160843145e-05,
+      "loss": 0.9616,
+      "step": 678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6781763277190615e-05,
+      "loss": 1.0103,
+      "step": 679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677176240246102e-05,
+      "loss": 0.9741,
+      "step": 680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6761749002740195e-05,
+      "loss": 0.9779,
+      "step": 681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6751723096548834e-05,
+      "loss": 0.9004,
+      "step": 682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674168470243076e-05,
+      "loss": 0.9575,
+      "step": 683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731633838952905e-05,
+      "loss": 0.9654,
+      "step": 684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672157052470525e-05,
+      "loss": 0.948,
+      "step": 685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671149477830081e-05,
+      "loss": 0.9752,
+      "step": 686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.67014066183756e-05,
+      "loss": 1.0078,
+      "step": 687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.9603,
+      "step": 688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6681193132621666e-05,
+      "loss": 0.9727,
+      "step": 689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671067844179625e-05,
+      "loss": 0.9364,
+      "step": 690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6660930216990112e-05,
+      "loss": 0.9507,
+      "step": 691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650780269803587e-05,
+      "loss": 0.9725,
+      "step": 692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640618021393305e-05,
+      "loss": 0.9234,
+      "step": 693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630443490555262e-05,
+      "loss": 0.9876,
+      "step": 694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6620256696108187e-05,
+      "loss": 0.8778,
+      "step": 695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610057656893483e-05,
+      "loss": 0.9502,
+      "step": 696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6599846391775197e-05,
+      "loss": 0.9265,
+      "step": 697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6589622919639996e-05,
+      "loss": 1.018,
+      "step": 698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657938725939713e-05,
+      "loss": 0.9362,
+      "step": 699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6569139429978373e-05,
+      "loss": 1.0856,
+      "step": 700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6558879450338023e-05,
+      "loss": 0.9787,
+      "step": 701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.8642,
+      "step": 702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653832311632206e-05,
+      "loss": 1.0131,
+      "step": 703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528026799967253e-05,
+      "loss": 0.9167,
+      "step": 704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517718409432405e-05,
+      "loss": 0.9866,
+      "step": 705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650739796378383e-05,
+      "loss": 0.9677,
+      "step": 706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6497065482110123e-05,
+      "loss": 0.8829,
+      "step": 707
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486720983522156e-05,
+      "loss": 0.9704,
+      "step": 708
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6476364487153024e-05,
+      "loss": 0.9651,
+      "step": 709
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6465996012157996e-05,
+      "loss": 0.9205,
+      "step": 710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645561557771453e-05,
+      "loss": 1.0237,
+      "step": 711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644522320302217e-05,
+      "loss": 0.951,
+      "step": 712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643481890730256e-05,
+      "loss": 0.9434,
+      "step": 713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424402709799404e-05,
+      "loss": 0.9903,
+      "step": 714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6413974629778396e-05,
+      "loss": 0.9018,
+      "step": 715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403534686527223e-05,
+      "loss": 0.888,
+      "step": 716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6393082899355516e-05,
+      "loss": 0.9526,
+      "step": 717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.63826192875948e-05,
+      "loss": 0.9421,
+      "step": 718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372143870598484e-05,
+      "loss": 0.9523,
+      "step": 719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63616566677418e-05,
+      "loss": 0.9914,
+      "step": 720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635115769842179e-05,
+      "loss": 0.9433,
+      "step": 721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634064698205725e-05,
+      "loss": 0.948,
+      "step": 722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6330124538088705e-05,
+      "loss": 0.9668,
+      "step": 723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631959038597838e-05,
+      "loss": 0.9267,
+      "step": 724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6309044545210144e-05,
+      "loss": 0.882,
+      "step": 725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629848703528949e-05,
+      "loss": 0.9424,
+      "step": 726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6287917875743495e-05,
+      "loss": 0.9159,
+      "step": 727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6277337086120783e-05,
+      "loss": 1.0005,
+      "step": 728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266744685991487e-05,
+      "loss": 0.9831,
+      "step": 729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6256140694947217e-05,
+      "loss": 0.9156,
+      "step": 730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245525132601016e-05,
+      "loss": 0.9892,
+      "step": 731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.8852,
+      "step": 732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622425937256199e-05,
+      "loss": 0.9117,
+      "step": 733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621360921420213e-05,
+      "loss": 1.0097,
+      "step": 734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.620294756320619e-05,
+      "loss": 0.9508,
+      "step": 735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6192274439293858e-05,
+      "loss": 0.97,
+      "step": 736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181589862206053e-05,
+      "loss": 0.9264,
+      "step": 737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6170893851704875e-05,
+      "loss": 0.9247,
+      "step": 738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6160186427573568e-05,
+      "loss": 0.9768,
+      "step": 739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614946760961649e-05,
+      "loss": 1.0188,
+      "step": 740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613873741765907e-05,
+      "loss": 1.057,
+      "step": 741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612799587154777e-05,
+      "loss": 1.0192,
+      "step": 742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117242991150064e-05,
+      "loss": 0.9408,
+      "step": 743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106478796354382e-05,
+      "loss": 0.937,
+      "step": 744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6095703307070084e-05,
+      "loss": 0.9818,
+      "step": 745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6084916543227413e-05,
+      "loss": 0.9642,
+      "step": 746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607411852477748e-05,
+      "loss": 0.9532,
+      "step": 747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606330927169219e-05,
+      "loss": 0.9383,
+      "step": 748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6052488803964256e-05,
+      "loss": 0.9175,
+      "step": 749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604165714160711e-05,
+      "loss": 0.9819,
+      "step": 750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6030814304654895e-05,
+      "loss": 0.9345,
+      "step": 751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6019960313162436e-05,
+      "loss": 0.9574,
+      "step": 752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600909518720517e-05,
+      "loss": 0.9621,
+      "step": 753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599821894687914e-05,
+      "loss": 0.9711,
+      "step": 754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598733161230094e-05,
+      "loss": 0.951,
+      "step": 755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597643320360769e-05,
+      "loss": 0.9182,
+      "step": 756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596552374095699e-05,
+      "loss": 0.9203,
+      "step": 757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595460324452688e-05,
+      "loss": 0.9466,
+      "step": 758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594367173451582e-05,
+      "loss": 0.8885,
+      "step": 759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5932729231142627e-05,
+      "loss": 0.9183,
+      "step": 760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592177575464646e-05,
+      "loss": 0.9102,
+      "step": 761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5910811325286768e-05,
+      "loss": 1.0198,
+      "step": 762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5899835963343265e-05,
+      "loss": 0.9349,
+      "step": 763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5888849689115882e-05,
+      "loss": 0.9457,
+      "step": 764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.9397,
+      "step": 765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586684448511008e-05,
+      "loss": 0.9904,
+      "step": 766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5855825596032287e-05,
+      "loss": 1.0333,
+      "step": 767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58447958760718e-05,
+      "loss": 1.0601,
+      "step": 768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583375534562908e-05,
+      "loss": 0.9194,
+      "step": 769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58227040251246e-05,
+      "loss": 0.9706,
+      "step": 770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581164193499879e-05,
+      "loss": 1.0188,
+      "step": 771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5800569095711983e-05,
+      "loss": 0.9263,
+      "step": 772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5789485527744407e-05,
+      "loss": 0.8956,
+      "step": 773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577839125159613e-05,
+      "loss": 0.9258,
+      "step": 774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5767286287787025e-05,
+      "loss": 0.945,
+      "step": 775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575617065685674e-05,
+      "loss": 0.9519,
+      "step": 776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5745044379364637e-05,
+      "loss": 1.0272,
+      "step": 777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733907475889782e-05,
+      "loss": 0.8945,
+      "step": 778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722759967030898e-05,
+      "loss": 0.9431,
+      "step": 779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5711601873406314e-05,
+      "loss": 0.9433,
+      "step": 780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5700433215653938e-05,
+      "loss": 0.9135,
+      "step": 781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5689254014431225e-05,
+      "loss": 0.9761,
+      "step": 782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5678064290415122e-05,
+      "loss": 0.8847,
+      "step": 783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566686406430204e-05,
+      "loss": 1.0627,
+      "step": 784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5655653356807826e-05,
+      "loss": 0.9397,
+      "step": 785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5644432188667695e-05,
+      "loss": 0.8947,
+      "step": 786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563320058063622e-05,
+      "loss": 1.0433,
+      "step": 787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5621958553487292e-05,
+      "loss": 0.9201,
+      "step": 788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5610706128014054e-05,
+      "loss": 0.874,
+      "step": 789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5599443325028896e-05,
+      "loss": 0.9473,
+      "step": 790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55881701653634e-05,
+      "loss": 0.9633,
+      "step": 791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5576886669868297e-05,
+      "loss": 0.9306,
+      "step": 792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565592859413442e-05,
+      "loss": 0.931,
+      "step": 793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555428875488776e-05,
+      "loss": 0.9501,
+      "step": 794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554297437719923e-05,
+      "loss": 0.7212,
+      "step": 795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553164974727482e-05,
+      "loss": 0.7314,
+      "step": 796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5520314886060467e-05,
+      "loss": 0.684,
+      "step": 797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5508969814521026e-05,
+      "loss": 0.7279,
+      "step": 798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549761455364024e-05,
+      "loss": 0.7354,
+      "step": 799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5486249124420702e-05,
+      "loss": 0.7226,
+      "step": 800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5474873547883803e-05,
+      "loss": 0.706,
+      "step": 801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5463487845069708e-05,
+      "loss": 0.678,
+      "step": 802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5452092037037317e-05,
+      "loss": 0.6841,
+      "step": 803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5440686144864207e-05,
+      "loss": 0.6594,
+      "step": 804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542927018964662e-05,
+      "loss": 0.7143,
+      "step": 805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.54178441924994e-05,
+      "loss": 0.6908,
+      "step": 806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.6102,
+      "step": 807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53949621569683e-05,
+      "loss": 0.6886,
+      "step": 808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5383506160906826e-05,
+      "loss": 0.7115,
+      "step": 809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5372040207560457e-05,
+      "loss": 0.7155,
+      "step": 810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5360564318136523e-05,
+      "loss": 0.7105,
+      "step": 811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349078513860728e-05,
+      "loss": 0.732,
+      "step": 812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5337582815977102e-05,
+      "loss": 0.6791,
+      "step": 813
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5326077245747998e-05,
+      "loss": 0.6891,
+      "step": 814
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314561824454014e-05,
+      "loss": 0.7763,
+      "step": 815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5303036573393964e-05,
+      "loss": 0.6054,
+      "step": 816
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529150151388485e-05,
+      "loss": 0.6794,
+      "step": 817
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527995666726182e-05,
+      "loss": 0.6833,
+      "step": 818
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5268402054878116e-05,
+      "loss": 0.6732,
+      "step": 819
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256837698105047e-05,
+      "loss": 0.6461,
+      "step": 820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245263618331944e-05,
+      "loss": 0.6487,
+      "step": 821
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5233679836966122e-05,
+      "loss": 0.6291,
+      "step": 822
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5222086375432837e-05,
+      "loss": 0.7249,
+      "step": 823
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210483255175255e-05,
+      "loss": 0.6289,
+      "step": 824
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198870497654397e-05,
+      "loss": 0.6835,
+      "step": 825
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187248124349125e-05,
+      "loss": 0.7099,
+      "step": 826
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175616156756072e-05,
+      "loss": 0.6865,
+      "step": 827
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163974616389621e-05,
+      "loss": 0.647,
+      "step": 828
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515232352478186e-05,
+      "loss": 0.6215,
+      "step": 829
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.514066290348255e-05,
+      "loss": 0.6791,
+      "step": 830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5128992774059063e-05,
+      "loss": 0.7277,
+      "step": 831
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5117313158096371e-05,
+      "loss": 0.6712,
+      "step": 832
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5105624077196988e-05,
+      "loss": 0.721,
+      "step": 833
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093925552980934e-05,
+      "loss": 0.6792,
+      "step": 834
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082217607085692e-05,
+      "loss": 0.6536,
+      "step": 835
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5070500261166176e-05,
+      "loss": 0.7053,
+      "step": 836
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5058773536894685e-05,
+      "loss": 0.6322,
+      "step": 837
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504703745596086e-05,
+      "loss": 0.6576,
+      "step": 838
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035292040071655e-05,
+      "loss": 0.6313,
+      "step": 839
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023537310951284e-05,
+      "loss": 0.6257,
+      "step": 840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501177329034119e-05,
+      "loss": 0.6438,
+      "step": 841
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6043,
+      "step": 842
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988217461703488e-05,
+      "loss": 0.6746,
+      "step": 843
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4976425697244531e-05,
+      "loss": 0.6574,
+      "step": 844
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4964624728433072e-05,
+      "loss": 0.7053,
+      "step": 845
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4952814577096073e-05,
+      "loss": 0.6729,
+      "step": 846
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4940995265077486e-05,
+      "loss": 0.6901,
+      "step": 847
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4929166814238211e-05,
+      "loss": 0.7331,
+      "step": 848
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491732924645604e-05,
+      "loss": 0.6998,
+      "step": 849
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490548258362563e-05,
+      "loss": 0.6326,
+      "step": 850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4893626847658466e-05,
+      "loss": 0.6779,
+      "step": 851
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4881762060482814e-05,
+      "loss": 0.6992,
+      "step": 852
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869888244043674e-05,
+      "loss": 0.6692,
+      "step": 853
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4858005420302752e-05,
+      "loss": 0.6722,
+      "step": 854
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4846113611238413e-05,
+      "loss": 0.6768,
+      "step": 855
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4834212838845639e-05,
+      "loss": 0.6839,
+      "step": 856
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4822303125135998e-05,
+      "loss": 0.6989,
+      "step": 857
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4810384492137582e-05,
+      "loss": 0.6642,
+      "step": 858
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4798456961894991e-05,
+      "loss": 0.6728,
+      "step": 859
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4786520556469284e-05,
+      "loss": 0.6987,
+      "step": 860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4774575297937921e-05,
+      "loss": 0.744,
+      "step": 861
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476262120839475e-05,
+      "loss": 0.6405,
+      "step": 862
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750658309949953e-05,
+      "loss": 0.6574,
+      "step": 863
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738686624729987e-05,
+      "loss": 0.69,
+      "step": 864
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4726706174877586e-05,
+      "loss": 0.6668,
+      "step": 865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471471698255168e-05,
+      "loss": 0.6986,
+      "step": 866
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4702719069927368e-05,
+      "loss": 0.6766,
+      "step": 867
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4690712459195888e-05,
+      "loss": 0.7236,
+      "step": 868
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4678697172564556e-05,
+      "loss": 0.6589,
+      "step": 869
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666673232256738e-05,
+      "loss": 0.6538,
+      "step": 870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465464066051181e-05,
+      "loss": 0.6794,
+      "step": 871
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642599479585106e-05,
+      "loss": 0.6184,
+      "step": 872
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630549711747888e-05,
+      "loss": 0.6757,
+      "step": 873
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618491379287301e-05,
+      "loss": 0.6971,
+      "step": 874
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4606424504506325e-05,
+      "loss": 0.6578,
+      "step": 875
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4594349109723744e-05,
+      "loss": 0.6426,
+      "step": 876
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4582265217274105e-05,
+      "loss": 0.6524,
+      "step": 877
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4570172849507664e-05,
+      "loss": 0.699,
+      "step": 878
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4558072028790356e-05,
+      "loss": 0.6399,
+      "step": 879
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454596277750375e-05,
+      "loss": 0.6781,
+      "step": 880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533845118045013e-05,
+      "loss": 0.6901,
+      "step": 881
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4521719072826858e-05,
+      "loss": 0.7126,
+      "step": 882
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4509584664277507e-05,
+      "loss": 0.7003,
+      "step": 883
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449744191484066e-05,
+      "loss": 0.8227,
+      "step": 884
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485290846975431e-05,
+      "loss": 0.6376,
+      "step": 885
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473131483156326e-05,
+      "loss": 0.6116,
+      "step": 886
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4460963845873203e-05,
+      "loss": 0.6856,
+      "step": 887
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4448787957631209e-05,
+      "loss": 0.7263,
+      "step": 888
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4436603840950763e-05,
+      "loss": 0.6551,
+      "step": 889
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4424411518367496e-05,
+      "loss": 0.7224,
+      "step": 890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4412211012432213e-05,
+      "loss": 0.7805,
+      "step": 891
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400002345710871e-05,
+      "loss": 0.6325,
+      "step": 892
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4387785540784504e-05,
+      "loss": 0.6974,
+      "step": 893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375560620249209e-05,
+      "loss": 0.6129,
+      "step": 894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4363327606716087e-05,
+      "loss": 0.6549,
+      "step": 895
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351086522811212e-05,
+      "loss": 0.6982,
+      "step": 896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.7061,
+      "step": 897
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4326580234465084e-05,
+      "loss": 0.7295,
+      "step": 898
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4314315075350442e-05,
+      "loss": 0.6955,
+      "step": 899
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430204193651719e-05,
+      "loss": 0.6525,
+      "step": 900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.428976084066561e-05,
+      "loss": 0.7106,
+      "step": 901
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427747181051071e-05,
+      "loss": 0.6228,
+      "step": 902
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4265174868782172e-05,
+      "loss": 0.7123,
+      "step": 903
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252870038224305e-05,
+      "loss": 0.7019,
+      "step": 904
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424055734159602e-05,
+      "loss": 0.6421,
+      "step": 905
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228236801670762e-05,
+      "loss": 0.7012,
+      "step": 906
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215908441236502e-05,
+      "loss": 0.6674,
+      "step": 907
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4203572283095657e-05,
+      "loss": 0.6904,
+      "step": 908
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4191228350065079e-05,
+      "loss": 0.6714,
+      "step": 909
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4178876664975994e-05,
+      "loss": 0.6766,
+      "step": 910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166517250673968e-05,
+      "loss": 0.6789,
+      "step": 911
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.6677,
+      "step": 912
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4141775325884804e-05,
+      "loss": 0.716,
+      "step": 913
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4129392861160102e-05,
+      "loss": 0.6927,
+      "step": 914
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4117002758747268e-05,
+      "loss": 0.7363,
+      "step": 915
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4104605041562918e-05,
+      "loss": 0.6829,
+      "step": 916
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4092199732537756e-05,
+      "loss": 0.6673,
+      "step": 917
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4079786854616537e-05,
+      "loss": 0.7361,
+      "step": 918
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.7206,
+      "step": 919
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4054938483934865e-05,
+      "loss": 0.6752,
+      "step": 920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4042503037133736e-05,
+      "loss": 0.6634,
+      "step": 921
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4030060113355118e-05,
+      "loss": 0.6675,
+      "step": 922
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4017609735613318e-05,
+      "loss": 0.6841,
+      "step": 923
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400515192693645e-05,
+      "loss": 0.6538,
+      "step": 924
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399268671036637e-05,
+      "loss": 0.6385,
+      "step": 925
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3980214108958626e-05,
+      "loss": 0.6855,
+      "step": 926
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3967734145782425e-05,
+      "loss": 0.6779,
+      "step": 927
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3955246843920607e-05,
+      "loss": 0.7307,
+      "step": 928
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3942752226469564e-05,
+      "loss": 0.7232,
+      "step": 929
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3930250316539237e-05,
+      "loss": 0.7077,
+      "step": 930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3917741137253042e-05,
+      "loss": 0.635,
+      "step": 931
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905224711747844e-05,
+      "loss": 0.7053,
+      "step": 932
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3892701063173917e-05,
+      "loss": 0.677,
+      "step": 933
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3880170214694886e-05,
+      "loss": 0.6245,
+      "step": 934
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867632189487693e-05,
+      "loss": 0.6997,
+      "step": 935
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3855087010742563e-05,
+      "loss": 0.7084,
+      "step": 936
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3842534701662947e-05,
+      "loss": 0.7482,
+      "step": 937
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3829975285465478e-05,
+      "loss": 0.6578,
+      "step": 938
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3817408785379944e-05,
+      "loss": 0.725,
+      "step": 939
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380483522464923e-05,
+      "loss": 0.7337,
+      "step": 940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3792254626529286e-05,
+      "loss": 0.6786,
+      "step": 941
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3779667014289067e-05,
+      "loss": 0.6593,
+      "step": 942
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3767072411210512e-05,
+      "loss": 0.6642,
+      "step": 943
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754470840588488e-05,
+      "loss": 0.685,
+      "step": 944
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741862325730739e-05,
+      "loss": 0.6701,
+      "step": 945
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372924688995787e-05,
+      "loss": 0.6132,
+      "step": 946
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3716624556603275e-05,
+      "loss": 0.7593,
+      "step": 947
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3703995349013113e-05,
+      "loss": 0.6376,
+      "step": 948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3691359290546248e-05,
+      "loss": 0.6936,
+      "step": 949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3678716404574222e-05,
+      "loss": 0.6836,
+      "step": 950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3666066714481205e-05,
+      "loss": 0.6351,
+      "step": 951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.6855,
+      "step": 952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640747015531757e-05,
+      "loss": 0.6602,
+      "step": 953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362807705350641e-05,
+      "loss": 0.6537,
+      "step": 954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3615400381022167e-05,
+      "loss": 0.6644,
+      "step": 955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3602717021525685e-05,
+      "loss": 0.6454,
+      "step": 956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3590026998475987e-05,
+      "loss": 0.6882,
+      "step": 957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3577330335344433e-05,
+      "loss": 0.6848,
+      "step": 958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3564627055614658e-05,
+      "loss": 0.6685,
+      "step": 959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355191718278253e-05,
+      "loss": 0.7289,
+      "step": 960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353920074035612e-05,
+      "loss": 0.6634,
+      "step": 961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3526477751855645e-05,
+      "loss": 0.6752,
+      "step": 962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3513748240813429e-05,
+      "loss": 0.6579,
+      "step": 963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3501012230773865e-05,
+      "loss": 0.726,
+      "step": 964
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3488269745293358e-05,
+      "loss": 0.665,
+      "step": 965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475520807940303e-05,
+      "loss": 0.7304,
+      "step": 966
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3462765442295018e-05,
+      "loss": 0.699,
+      "step": 967
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3450003671949707e-05,
+      "loss": 0.6871,
+      "step": 968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3437235520508434e-05,
+      "loss": 0.6858,
+      "step": 969
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3424461011587058e-05,
+      "loss": 0.7196,
+      "step": 970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341168016881319e-05,
+      "loss": 0.6594,
+      "step": 971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3398893015826166e-05,
+      "loss": 0.6844,
+      "step": 972
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3386099576276995e-05,
+      "loss": 0.6634,
+      "step": 973
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3373299873828303e-05,
+      "loss": 0.6097,
+      "step": 974
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3360493932154301e-05,
+      "loss": 0.6566,
+      "step": 975
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3347681774940756e-05,
+      "loss": 0.648,
+      "step": 976
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3334863425884907e-05,
+      "loss": 0.6526,
+      "step": 977
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322038908695466e-05,
+      "loss": 0.6868,
+      "step": 978
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309208247092546e-05,
+      "loss": 0.7064,
+      "step": 979
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3296371464807618e-05,
+      "loss": 0.6615,
+      "step": 980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3283528585583484e-05,
+      "loss": 0.6969,
+      "step": 981
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3270679633174219e-05,
+      "loss": 0.6803,
+      "step": 982
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3257824631345126e-05,
+      "loss": 0.6952,
+      "step": 983
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244963603872707e-05,
+      "loss": 0.6623,
+      "step": 984
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3232096574544602e-05,
+      "loss": 0.661,
+      "step": 985
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219223567159551e-05,
+      "loss": 0.5954,
+      "step": 986
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3206344605527355e-05,
+      "loss": 0.6665,
+      "step": 987
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193459713468827e-05,
+      "loss": 0.6922,
+      "step": 988
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3180568914815752e-05,
+      "loss": 0.6976,
+      "step": 989
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3167672233410826e-05,
+      "loss": 0.6623,
+      "step": 990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3154769693107645e-05,
+      "loss": 0.7077,
+      "step": 991
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3141861317770628e-05,
+      "loss": 0.6812,
+      "step": 992
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312894713127499e-05,
+      "loss": 0.7365,
+      "step": 993
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3116027157506696e-05,
+      "loss": 0.6025,
+      "step": 994
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3103101420362414e-05,
+      "loss": 0.6775,
+      "step": 995
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.6485,
+      "step": 996
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.307723275158582e-05,
+      "loss": 0.6433,
+      "step": 997
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3064289867799962e-05,
+      "loss": 0.6639,
+      "step": 998
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3051341316330946e-05,
+      "loss": 0.6693,
+      "step": 999
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3038387121128298e-05,
+      "loss": 0.7057,
+      "step": 1000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3025427306151976e-05,
+      "loss": 0.6729,
+      "step": 1001
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3012461895372343e-05,
+      "loss": 0.6997,
+      "step": 1002
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2999490912770108e-05,
+      "loss": 0.6814,
+      "step": 1003
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2986514382336281e-05,
+      "loss": 0.6811,
+      "step": 1004
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973532328072138e-05,
+      "loss": 0.6854,
+      "step": 1005
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2960544773989172e-05,
+      "loss": 0.7497,
+      "step": 1006
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2947551744109044e-05,
+      "loss": 0.6745,
+      "step": 1007
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934553262463548e-05,
+      "loss": 0.6779,
+      "step": 1008
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292154935309456e-05,
+      "loss": 0.7395,
+      "step": 1009
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908540040053992e-05,
+      "loss": 0.7092,
+      "step": 1010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2895525347403757e-05,
+      "loss": 0.6859,
+      "step": 1011
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2882505299215711e-05,
+      "loss": 0.6412,
+      "step": 1012
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.286947991957162e-05,
+      "loss": 0.6955,
+      "step": 1013
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285644923256311e-05,
+      "loss": 0.7113,
+      "step": 1014
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2843413262291621e-05,
+      "loss": 0.6689,
+      "step": 1015
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283037203286837e-05,
+      "loss": 0.7193,
+      "step": 1016
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.6723,
+      "step": 1017
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2804273893060028e-05,
+      "loss": 0.6405,
+      "step": 1018
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2791217030945824e-05,
+      "loss": 0.652,
+      "step": 1019
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277815500622154e-05,
+      "loss": 0.6962,
+      "step": 1020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276508784304658e-05,
+      "loss": 0.7274,
+      "step": 1021
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2752015565589852e-05,
+      "loss": 0.6515,
+      "step": 1022
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2738938198029725e-05,
+      "loss": 0.6786,
+      "step": 1023
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2725855764553981e-05,
+      "loss": 0.6621,
+      "step": 1024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2712768289359767e-05,
+      "loss": 0.7016,
+      "step": 1025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269967579665357e-05,
+      "loss": 0.6618,
+      "step": 1026
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268657831065114e-05,
+      "loss": 0.7137,
+      "step": 1027
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673475855577473e-05,
+      "loss": 0.6827,
+      "step": 1028
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2660368455666752e-05,
+      "loss": 0.6838,
+      "step": 1029
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647256135162307e-05,
+      "loss": 0.701,
+      "step": 1030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634138918316567e-05,
+      "loss": 0.6991,
+      "step": 1031
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2621016829391022e-05,
+      "loss": 0.6148,
+      "step": 1032
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607889892656171e-05,
+      "loss": 0.6552,
+      "step": 1033
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2594758132391473e-05,
+      "loss": 0.6856,
+      "step": 1034
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2581621572885324e-05,
+      "loss": 0.6725,
+      "step": 1035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2568480238434977e-05,
+      "loss": 0.6935,
+      "step": 1036
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2555334153346533e-05,
+      "loss": 0.6536,
+      "step": 1037
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542183341934873e-05,
+      "loss": 0.6979,
+      "step": 1038
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2529027828523614e-05,
+      "loss": 0.672,
+      "step": 1039
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515867637445088e-05,
+      "loss": 0.6955,
+      "step": 1040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502702793040257e-05,
+      "loss": 0.6881,
+      "step": 1041
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2489533319658703e-05,
+      "loss": 0.7223,
+      "step": 1042
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2476359241658562e-05,
+      "loss": 0.6372,
+      "step": 1043
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.24631805834065e-05,
+      "loss": 0.6158,
+      "step": 1044
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244999736927764e-05,
+      "loss": 0.6109,
+      "step": 1045
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2436809623655535e-05,
+      "loss": 0.6855,
+      "step": 1046
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2423617370932127e-05,
+      "loss": 0.681,
+      "step": 1047
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410420635507682e-05,
+      "loss": 0.6707,
+      "step": 1048
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2397219441790768e-05,
+      "loss": 0.6883,
+      "step": 1049
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2384013814198197e-05,
+      "loss": 0.6478,
+      "step": 1050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2370803777154976e-05,
+      "loss": 0.6429,
+      "step": 1051
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357589355094275e-05,
+      "loss": 0.7186,
+      "step": 1052
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2344370572457368e-05,
+      "loss": 0.7129,
+      "step": 1053
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2331147453693594e-05,
+      "loss": 0.6663,
+      "step": 1054
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2317920023260323e-05,
+      "loss": 0.6325,
+      "step": 1055
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2304688305622889e-05,
+      "loss": 0.7273,
+      "step": 1056
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291452325254556e-05,
+      "loss": 0.7508,
+      "step": 1057
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2278212106636478e-05,
+      "loss": 0.6062,
+      "step": 1058
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2264967674257647e-05,
+      "loss": 0.6681,
+      "step": 1059
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2251719052614841e-05,
+      "loss": 0.6414,
+      "step": 1060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2238466266212594e-05,
+      "loss": 0.6821,
+      "step": 1061
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.6357,
+      "step": 1062
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2211948297186384e-05,
+      "loss": 0.7001,
+      "step": 1063
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198683163609817e-05,
+      "loss": 0.6735,
+      "step": 1064
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218541396336852e-05,
+      "loss": 0.6966,
+      "step": 1065
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217214072100508e-05,
+      "loss": 0.7338,
+      "step": 1066
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2158863461069573e-05,
+      "loss": 0.6841,
+      "step": 1067
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2145582208119497e-05,
+      "loss": 0.7123,
+      "step": 1068
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213229698671974e-05,
+      "loss": 0.7284,
+      "step": 1069
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2119007821442533e-05,
+      "loss": 0.6856,
+      "step": 1070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2105714736867391e-05,
+      "loss": 0.6671,
+      "step": 1071
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2092417757581085e-05,
+      "loss": 0.6751,
+      "step": 1072
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 0.6834,
+      "step": 1073
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2065812213258051e-05,
+      "loss": 0.649,
+      "step": 1074
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2052503697430691e-05,
+      "loss": 0.6553,
+      "step": 1075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2039191385310831e-05,
+      "loss": 0.6736,
+      "step": 1076
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2025875301520811e-05,
+      "loss": 0.7656,
+      "step": 1077
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012555470689935e-05,
+      "loss": 0.6837,
+      "step": 1078
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199923191745444e-05,
+      "loss": 0.756,
+      "step": 1079
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1985904666457455e-05,
+      "loss": 0.6466,
+      "step": 1080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1972573742348947e-05,
+      "loss": 0.6532,
+      "step": 1081
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1959239169785668e-05,
+      "loss": 0.7052,
+      "step": 1082
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1945900973431127e-05,
+      "loss": 0.7102,
+      "step": 1083
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1932559177955533e-05,
+      "loss": 0.6537,
+      "step": 1084
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919213808035755e-05,
+      "loss": 0.6751,
+      "step": 1085
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1905864888355264e-05,
+      "loss": 0.6818,
+      "step": 1086
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1892512443604103e-05,
+      "loss": 0.733,
+      "step": 1087
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1879156498478838e-05,
+      "loss": 0.6979,
+      "step": 1088
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1865797077682509e-05,
+      "loss": 0.664,
+      "step": 1089
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185243420592457e-05,
+      "loss": 0.7206,
+      "step": 1090
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1839067907920879e-05,
+      "loss": 0.642,
+      "step": 1091
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.182569820839362e-05,
+      "loss": 0.6945,
+      "step": 1092
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812325132071265e-05,
+      "loss": 0.6545,
+      "step": 1093
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.179894870368854e-05,
+      "loss": 0.7185,
+      "step": 1094
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785568947986368e-05,
+      "loss": 0.6783,
+      "step": 1095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1772185889711825e-05,
+      "loss": 0.6688,
+      "step": 1096
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1758799553618094e-05,
+      "loss": 0.6292,
+      "step": 1097
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1745409964464425e-05,
+      "loss": 0.6851,
+      "step": 1098
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1732017147016082e-05,
+      "loss": 0.7023,
+      "step": 1099
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.17186211260443e-05,
+      "loss": 0.6909,
+      "step": 1100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170522192632624e-05,
+      "loss": 0.697,
+      "step": 1101
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691819572644941e-05,
+      "loss": 0.6669,
+      "step": 1102
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1678414089789278e-05,
+      "loss": 0.6219,
+      "step": 1103
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1665005502553912e-05,
+      "loss": 0.676,
+      "step": 1104
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1651593835739245e-05,
+      "loss": 0.6917,
+      "step": 1105
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1638179114151378e-05,
+      "loss": 0.6828,
+      "step": 1106
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1624761362602061e-05,
+      "loss": 0.6151,
+      "step": 1107
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1611340605908643e-05,
+      "loss": 0.7257,
+      "step": 1108
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597916868894038e-05,
+      "loss": 0.6779,
+      "step": 1109
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1584490176386671e-05,
+      "loss": 0.6873,
+      "step": 1110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1571060553220432e-05,
+      "loss": 0.6783,
+      "step": 1111
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155762802423463e-05,
+      "loss": 0.6771,
+      "step": 1112
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1544192614273957e-05,
+      "loss": 0.6807,
+      "step": 1113
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1530754348188418e-05,
+      "loss": 0.6203,
+      "step": 1114
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1517313250833318e-05,
+      "loss": 0.6891,
+      "step": 1115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1503869347069186e-05,
+      "loss": 0.6687,
+      "step": 1116
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1490422661761744e-05,
+      "loss": 0.6414,
+      "step": 1117
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476973219781868e-05,
+      "loss": 0.6656,
+      "step": 1118
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1463521046005524e-05,
+      "loss": 0.648,
+      "step": 1119
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1450066165313724e-05,
+      "loss": 0.7562,
+      "step": 1120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1436608602592505e-05,
+      "loss": 0.7681,
+      "step": 1121
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.6976,
+      "step": 1122
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1409685530630667e-05,
+      "loss": 0.6729,
+      "step": 1123
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1396220071186718e-05,
+      "loss": 0.6805,
+      "step": 1124
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1382752029306606e-05,
+      "loss": 0.6403,
+      "step": 1125
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136928142990069e-05,
+      "loss": 0.6624,
+      "step": 1126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1355808297884078e-05,
+      "loss": 0.6928,
+      "step": 1127
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342332658176556e-05,
+      "loss": 0.5906,
+      "step": 1128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1328854535702542e-05,
+      "loss": 0.6484,
+      "step": 1129
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1315373955391051e-05,
+      "loss": 0.7119,
+      "step": 1130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1301890942175649e-05,
+      "loss": 0.6709,
+      "step": 1131
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128840552099439e-05,
+      "loss": 0.7097,
+      "step": 1132
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274917716789793e-05,
+      "loss": 0.6109,
+      "step": 1133
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.126142755450878e-05,
+      "loss": 0.7222,
+      "step": 1134
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124793505910263e-05,
+      "loss": 0.6624,
+      "step": 1135
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234440255526948e-05,
+      "loss": 0.6695,
+      "step": 1136
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12209431687416e-05,
+      "loss": 0.7271,
+      "step": 1137
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207443823710676e-05,
+      "loss": 0.6289,
+      "step": 1138
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193942245402443e-05,
+      "loss": 0.7308,
+      "step": 1139
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1180438458789305e-05,
+      "loss": 0.6895,
+      "step": 1140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116693248884774e-05,
+      "loss": 0.6967,
+      "step": 1141
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1153424360558268e-05,
+      "loss": 0.6882,
+      "step": 1142
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1139914098905406e-05,
+      "loss": 0.6727,
+      "step": 1143
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1126401728877616e-05,
+      "loss": 0.6655,
+      "step": 1144
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1112887275467248e-05,
+      "loss": 0.7333,
+      "step": 1145
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1099370763670523e-05,
+      "loss": 0.6769,
+      "step": 1146
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1085852218487453e-05,
+      "loss": 0.6401,
+      "step": 1147
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1072331664921826e-05,
+      "loss": 0.689,
+      "step": 1148
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058809127981133e-05,
+      "loss": 0.7531,
+      "step": 1149
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.7504,
+      "step": 1150
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1031758204022824e-05,
+      "loss": 0.7011,
+      "step": 1151
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1018229867038358e-05,
+      "loss": 0.6603,
+      "step": 1152
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1004699646745023e-05,
+      "loss": 0.6771,
+      "step": 1153
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0991167568168202e-05,
+      "loss": 0.714,
+      "step": 1154
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0977633656336706e-05,
+      "loss": 0.6607,
+      "step": 1155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.096409793628273e-05,
+      "loss": 0.6362,
+      "step": 1156
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0950560433041825e-05,
+      "loss": 0.7207,
+      "step": 1157
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937021171652842e-05,
+      "loss": 0.6502,
+      "step": 1158
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0923480177157871e-05,
+      "loss": 0.7322,
+      "step": 1159
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0909937474602215e-05,
+      "loss": 0.6823,
+      "step": 1160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0896393089034336e-05,
+      "loss": 0.68,
+      "step": 1161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0882847045505809e-05,
+      "loss": 0.6787,
+      "step": 1162
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.086929936907128e-05,
+      "loss": 0.6188,
+      "step": 1163
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.08557500847884e-05,
+      "loss": 0.6899,
+      "step": 1164
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0842199217717808e-05,
+      "loss": 0.7358,
+      "step": 1165
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0828646792923074e-05,
+      "loss": 0.6822,
+      "step": 1166
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0815092835470633e-05,
+      "loss": 0.6679,
+      "step": 1167
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0801537370429767e-05,
+      "loss": 0.7089,
+      "step": 1168
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0787980422872543e-05,
+      "loss": 0.7237,
+      "step": 1169
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077442201787377e-05,
+      "loss": 0.6331,
+      "step": 1170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076086218051095e-05,
+      "loss": 0.7445,
+      "step": 1171
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.6925,
+      "step": 1172
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0733738309016402e-05,
+      "loss": 0.6357,
+      "step": 1173
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072017432505274e-05,
+      "loss": 0.7069,
+      "step": 1174
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0706609009061082e-05,
+      "loss": 0.7105,
+      "step": 1175
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0693042386131713e-05,
+      "loss": 0.7134,
+      "step": 1176
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0679474481357342e-05,
+      "loss": 0.6632,
+      "step": 1177
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066590531983304e-05,
+      "loss": 0.6837,
+      "step": 1178
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0652334926656209e-05,
+      "loss": 0.7547,
+      "step": 1179
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638763326926524e-05,
+      "loss": 0.6425,
+      "step": 1180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.06251905457459e-05,
+      "loss": 0.649,
+      "step": 1181
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0611616608218429e-05,
+      "loss": 0.6579,
+      "step": 1182
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598041539450344e-05,
+      "loss": 0.7468,
+      "step": 1183
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058446536454997e-05,
+      "loss": 0.6516,
+      "step": 1184
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0570888108627682e-05,
+      "loss": 0.735,
+      "step": 1185
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0557309796795848e-05,
+      "loss": 0.761,
+      "step": 1186
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0543730454168794e-05,
+      "loss": 0.6928,
+      "step": 1187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0530150105862748e-05,
+      "loss": 0.7577,
+      "step": 1188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0516568776995809e-05,
+      "loss": 0.686,
+      "step": 1189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0502986492687872e-05,
+      "loss": 0.6566,
+      "step": 1190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0489403278060613e-05,
+      "loss": 0.6918,
+      "step": 1191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0475819158237426e-05,
+      "loss": 0.6991,
+      "step": 1192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0462234158343374e-05,
+      "loss": 0.6313,
+      "step": 1193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044864830350515e-05,
+      "loss": 0.6661,
+      "step": 1194
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0435061618851034e-05,
+      "loss": 0.6755,
+      "step": 1195
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042147412951084e-05,
+      "loss": 0.6793,
+      "step": 1196
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0407885860615859e-05,
+      "loss": 0.6945,
+      "step": 1197
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0394296837298832e-05,
+      "loss": 0.6902,
+      "step": 1198
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380707084693902e-05,
+      "loss": 0.6672,
+      "step": 1199
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0367116627936549e-05,
+      "loss": 0.7167,
+      "step": 1200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035352549216356e-05,
+      "loss": 0.6633,
+      "step": 1201
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339933702512978e-05,
+      "loss": 0.6658,
+      "step": 1202
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0326341284124061e-05,
+      "loss": 0.6676,
+      "step": 1203
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0312748262137217e-05,
+      "loss": 0.7418,
+      "step": 1204
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0299154661693987e-05,
+      "loss": 0.6116,
+      "step": 1205
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0285560507936962e-05,
+      "loss": 0.6884,
+      "step": 1206
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0271965826009778e-05,
+      "loss": 0.6803,
+      "step": 1207
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0258370641057031e-05,
+      "loss": 0.7002,
+      "step": 1208
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0244774978224254e-05,
+      "loss": 0.7043,
+      "step": 1209
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0231178862657862e-05,
+      "loss": 0.6697,
+      "step": 1210
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.021758231950512e-05,
+      "loss": 0.7034,
+      "step": 1211
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0203985373914056e-05,
+      "loss": 0.7343,
+      "step": 1212
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0190388051033466e-05,
+      "loss": 0.5975,
+      "step": 1213
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176790376012843e-05,
+      "loss": 0.6847,
+      "step": 1214
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.016319237400232e-05,
+      "loss": 0.7021,
+      "step": 1215
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0149594070152638e-05,
+      "loss": 0.7213,
+      "step": 1216
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01359954896151e-05,
+      "loss": 0.6457,
+      "step": 1217
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0122396657541522e-05,
+      "loss": 0.7275,
+      "step": 1218
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0108797599084176e-05,
+      "loss": 0.6945,
+      "step": 1219
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0095198339395769e-05,
+      "loss": 0.7165,
+      "step": 1220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0081598903629359e-05,
+      "loss": 0.7053,
+      "step": 1221
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0067999316938348e-05,
+      "loss": 0.6489,
+      "step": 1222
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0054399604476408e-05,
+      "loss": 0.685,
+      "step": 1223
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0040799791397444e-05,
+      "loss": 0.6287,
+      "step": 1224
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0027199902855553e-05,
+      "loss": 0.6538,
+      "step": 1225
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013599964004966e-05,
+      "loss": 0.6572,
+      "step": 1226
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.7011,
+      "step": 1227
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98640003599504e-06,
+      "loss": 0.6891,
+      "step": 1228
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972800097144449e-06,
+      "loss": 0.6857,
+      "step": 1229
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95920020860256e-06,
+      "loss": 0.7006,
+      "step": 1230
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.945600395523597e-06,
+      "loss": 0.6833,
+      "step": 1231
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.932000683061654e-06,
+      "loss": 0.6981,
+      "step": 1232
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918401096370644e-06,
+      "loss": 0.6141,
+      "step": 1233
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.904801660604234e-06,
+      "loss": 0.708,
+      "step": 1234
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891202400915822e-06,
+      "loss": 0.6731,
+      "step": 1235
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877603342458483e-06,
+      "loss": 0.6491,
+      "step": 1236
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864004510384903e-06,
+      "loss": 0.6837,
+      "step": 1237
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.850405929847367e-06,
+      "loss": 0.6843,
+      "step": 1238
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.836807625997683e-06,
+      "loss": 0.6555,
+      "step": 1239
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.82320962398716e-06,
+      "loss": 0.7153,
+      "step": 1240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.809611948966534e-06,
+      "loss": 0.6558,
+      "step": 1241
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79601462608595e-06,
+      "loss": 0.6617,
+      "step": 1242
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.782417680494887e-06,
+      "loss": 0.6621,
+      "step": 1243
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76882113734214e-06,
+      "loss": 0.6428,
+      "step": 1244
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75522502177575e-06,
+      "loss": 0.6723,
+      "step": 1245
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74162935894297e-06,
+      "loss": 0.7024,
+      "step": 1246
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.728034173990224e-06,
+      "loss": 0.6595,
+      "step": 1247
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71443949206304e-06,
+      "loss": 0.6709,
+      "step": 1248
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.700845338306018e-06,
+      "loss": 0.6664,
+      "step": 1249
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.687251737862785e-06,
+      "loss": 0.7001,
+      "step": 1250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67365871587594e-06,
+      "loss": 0.6334,
+      "step": 1251
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.660066297487024e-06,
+      "loss": 0.6294,
+      "step": 1252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.646474507836442e-06,
+      "loss": 0.6792,
+      "step": 1253
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.632883372063458e-06,
+      "loss": 0.6829,
+      "step": 1254
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619292915306103e-06,
+      "loss": 0.6436,
+      "step": 1255
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.605703162701172e-06,
+      "loss": 0.6775,
+      "step": 1256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.592114139384146e-06,
+      "loss": 0.6801,
+      "step": 1257
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.578525870489163e-06,
+      "loss": 0.6599,
+      "step": 1258
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.564938381148966e-06,
+      "loss": 0.6491,
+      "step": 1259
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.551351696494854e-06,
+      "loss": 0.6976,
+      "step": 1260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.537765841656632e-06,
+      "loss": 0.6734,
+      "step": 1261
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524180841762577e-06,
+      "loss": 0.7301,
+      "step": 1262
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51059672193939e-06,
+      "loss": 0.7687,
+      "step": 1263
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49701350731213e-06,
+      "loss": 0.7074,
+      "step": 1264
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.483431223004193e-06,
+      "loss": 0.7367,
+      "step": 1265
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.469849894137254e-06,
+      "loss": 0.6635,
+      "step": 1266
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45626954583121e-06,
+      "loss": 0.6348,
+      "step": 1267
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.442690203204156e-06,
+      "loss": 0.6415,
+      "step": 1268
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42911189137232e-06,
+      "loss": 0.6562,
+      "step": 1269
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415534635450032e-06,
+      "loss": 0.6642,
+      "step": 1270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401958460549658e-06,
+      "loss": 0.6716,
+      "step": 1271
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.388383391781576e-06,
+      "loss": 0.6506,
+      "step": 1272
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.374809454254102e-06,
+      "loss": 0.6686,
+      "step": 1273
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361236673073479e-06,
+      "loss": 0.6894,
+      "step": 1274
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.347665073343795e-06,
+      "loss": 0.6633,
+      "step": 1275
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.334094680166962e-06,
+      "loss": 0.6741,
+      "step": 1276
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.320525518642661e-06,
+      "loss": 0.7106,
+      "step": 1277
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.306957613868292e-06,
+      "loss": 0.675,
+      "step": 1278
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293390990938921e-06,
+      "loss": 0.7012,
+      "step": 1279
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279825674947263e-06,
+      "loss": 0.6568,
+      "step": 1280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266261690983601e-06,
+      "loss": 0.6716,
+      "step": 1281
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 0.6561,
+      "step": 1282
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.239137819489047e-06,
+      "loss": 0.7007,
+      "step": 1283
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225577982126234e-06,
+      "loss": 0.6767,
+      "step": 1284
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.212019577127462e-06,
+      "loss": 0.601,
+      "step": 1285
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198462629570236e-06,
+      "loss": 0.6375,
+      "step": 1286
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184907164529369e-06,
+      "loss": 0.672,
+      "step": 1287
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17135320707693e-06,
+      "loss": 0.7109,
+      "step": 1288
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.157800782282192e-06,
+      "loss": 0.6617,
+      "step": 1289
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.144249915211605e-06,
+      "loss": 0.6977,
+      "step": 1290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.130700630928726e-06,
+      "loss": 0.6503,
+      "step": 1291
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.117152954494195e-06,
+      "loss": 0.6483,
+      "step": 1292
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103606910965666e-06,
+      "loss": 0.6055,
+      "step": 1293
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.090062525397787e-06,
+      "loss": 0.6918,
+      "step": 1294
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07651982284213e-06,
+      "loss": 0.619,
+      "step": 1295
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06297882834716e-06,
+      "loss": 0.6847,
+      "step": 1296
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049439566958176e-06,
+      "loss": 0.6002,
+      "step": 1297
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035902063717272e-06,
+      "loss": 0.6557,
+      "step": 1298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.022366343663299e-06,
+      "loss": 0.6053,
+      "step": 1299
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.0088324318318e-06,
+      "loss": 0.6704,
+      "step": 1300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995300353254977e-06,
+      "loss": 0.6568,
+      "step": 1301
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.981770132961649e-06,
+      "loss": 0.6181,
+      "step": 1302
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96824179597718e-06,
+      "loss": 0.7049,
+      "step": 1303
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.7194,
+      "step": 1304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94119087201887e-06,
+      "loss": 0.6663,
+      "step": 1305
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.927668335078175e-06,
+      "loss": 0.6826,
+      "step": 1306
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.914147781512547e-06,
+      "loss": 0.6206,
+      "step": 1307
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.900629236329482e-06,
+      "loss": 0.732,
+      "step": 1308
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887112724532755e-06,
+      "loss": 0.7113,
+      "step": 1309
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.873598271122387e-06,
+      "loss": 0.7261,
+      "step": 1310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.860085901094595e-06,
+      "loss": 0.6432,
+      "step": 1311
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.846575639441732e-06,
+      "loss": 0.6799,
+      "step": 1312
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83306751115226e-06,
+      "loss": 0.6766,
+      "step": 1313
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819561541210698e-06,
+      "loss": 0.6739,
+      "step": 1314
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806057754597559e-06,
+      "loss": 0.6716,
+      "step": 1315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.792556176289326e-06,
+      "loss": 0.6698,
+      "step": 1316
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.779056831258402e-06,
+      "loss": 0.6586,
+      "step": 1317
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765559744473054e-06,
+      "loss": 0.6065,
+      "step": 1318
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75206494089737e-06,
+      "loss": 0.6771,
+      "step": 1319
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.738572445491225e-06,
+      "loss": 0.7097,
+      "step": 1320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72508228321021e-06,
+      "loss": 0.7034,
+      "step": 1321
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.711594479005614e-06,
+      "loss": 0.6138,
+      "step": 1322
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698109057824354e-06,
+      "loss": 0.6342,
+      "step": 1323
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.684626044608949e-06,
+      "loss": 0.7309,
+      "step": 1324
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67114546429746e-06,
+      "loss": 0.7281,
+      "step": 1325
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657667341823449e-06,
+      "loss": 0.7636,
+      "step": 1326
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644191702115923e-06,
+      "loss": 0.7073,
+      "step": 1327
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63071857009931e-06,
+      "loss": 0.6625,
+      "step": 1328
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6172479706934e-06,
+      "loss": 0.6618,
+      "step": 1329
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.603779928813283e-06,
+      "loss": 0.6698,
+      "step": 1330
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.590314469369335e-06,
+      "loss": 0.71,
+      "step": 1331
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.6689,
+      "step": 1332
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.563391397407498e-06,
+      "loss": 0.6445,
+      "step": 1333
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.549933834686278e-06,
+      "loss": 0.6943,
+      "step": 1334
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53647895399448e-06,
+      "loss": 0.6816,
+      "step": 1335
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.523026780218134e-06,
+      "loss": 0.6888,
+      "step": 1336
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.509577338238255e-06,
+      "loss": 0.6776,
+      "step": 1337
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.496130652930818e-06,
+      "loss": 0.7002,
+      "step": 1338
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.482686749166685e-06,
+      "loss": 0.6518,
+      "step": 1339
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.469245651811584e-06,
+      "loss": 0.6386,
+      "step": 1340
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.455807385726047e-06,
+      "loss": 0.6853,
+      "step": 1341
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.442371975765368e-06,
+      "loss": 0.6767,
+      "step": 1342
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42893944677957e-06,
+      "loss": 0.6775,
+      "step": 1343
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.415509823613332e-06,
+      "loss": 0.7005,
+      "step": 1344
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.402083131105966e-06,
+      "loss": 0.6117,
+      "step": 1345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.388659394091362e-06,
+      "loss": 0.6205,
+      "step": 1346
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.375238637397942e-06,
+      "loss": 0.6848,
+      "step": 1347
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.361820885848623e-06,
+      "loss": 0.702,
+      "step": 1348
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.348406164260755e-06,
+      "loss": 0.6675,
+      "step": 1349
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.334994497446091e-06,
+      "loss": 0.7143,
+      "step": 1350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.321585910210725e-06,
+      "loss": 0.6801,
+      "step": 1351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308180427355062e-06,
+      "loss": 0.624,
+      "step": 1352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.294778073673762e-06,
+      "loss": 0.7046,
+      "step": 1353
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.281378873955701e-06,
+      "loss": 0.7257,
+      "step": 1354
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.26798285298392e-06,
+      "loss": 0.6934,
+      "step": 1355
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25459003553558e-06,
+      "loss": 0.7874,
+      "step": 1356
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24120044638191e-06,
+      "loss": 0.617,
+      "step": 1357
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.227814110288178e-06,
+      "loss": 0.6183,
+      "step": 1358
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214431052013636e-06,
+      "loss": 0.6513,
+      "step": 1359
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.201051296311462e-06,
+      "loss": 0.5937,
+      "step": 1360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18767486792874e-06,
+      "loss": 0.6711,
+      "step": 1361
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.174301791606384e-06,
+      "loss": 0.6485,
+      "step": 1362
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.160932092079125e-06,
+      "loss": 0.6412,
+      "step": 1363
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147565794075433e-06,
+      "loss": 0.7094,
+      "step": 1364
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.134202922317496e-06,
+      "loss": 0.6159,
+      "step": 1365
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.120843501521164e-06,
+      "loss": 0.7157,
+      "step": 1366
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.107487556395902e-06,
+      "loss": 0.6575,
+      "step": 1367
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.094135111644741e-06,
+      "loss": 0.661,
+      "step": 1368
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08078619196425e-06,
+      "loss": 0.6331,
+      "step": 1369
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06744082204447e-06,
+      "loss": 0.6273,
+      "step": 1370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.054099026568874e-06,
+      "loss": 0.669,
+      "step": 1371
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.040760830214334e-06,
+      "loss": 0.6733,
+      "step": 1372
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.02742625765106e-06,
+      "loss": 0.6035,
+      "step": 1373
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.014095333542548e-06,
+      "loss": 0.65,
+      "step": 1374
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.000768082545564e-06,
+      "loss": 0.6457,
+      "step": 1375
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.987444529310068e-06,
+      "loss": 0.6492,
+      "step": 1376
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.974124698479192e-06,
+      "loss": 0.7322,
+      "step": 1377
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.960808614689169e-06,
+      "loss": 0.6441,
+      "step": 1378
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.947496302569315e-06,
+      "loss": 0.6323,
+      "step": 1379
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.934187786741956e-06,
+      "loss": 0.6487,
+      "step": 1380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 0.622,
+      "step": 1381
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907582242418916e-06,
+      "loss": 0.6773,
+      "step": 1382
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.894285263132612e-06,
+      "loss": 0.6892,
+      "step": 1383
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88099217855747e-06,
+      "loss": 0.6825,
+      "step": 1384
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.867703013280263e-06,
+      "loss": 0.6157,
+      "step": 1385
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.854417791880508e-06,
+      "loss": 0.6651,
+      "step": 1386
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.841136538930432e-06,
+      "loss": 0.6556,
+      "step": 1387
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827859278994924e-06,
+      "loss": 0.6927,
+      "step": 1388
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.814586036631485e-06,
+      "loss": 0.6924,
+      "step": 1389
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.801316836390183e-06,
+      "loss": 0.6385,
+      "step": 1390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78805170281362e-06,
+      "loss": 0.6459,
+      "step": 1391
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.6726,
+      "step": 1392
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.761533733787409e-06,
+      "loss": 0.713,
+      "step": 1393
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.748280947385162e-06,
+      "loss": 0.6745,
+      "step": 1394
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.735032325742355e-06,
+      "loss": 0.5446,
+      "step": 1395
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.721787893363524e-06,
+      "loss": 0.6661,
+      "step": 1396
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708547674745447e-06,
+      "loss": 0.6578,
+      "step": 1397
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.695311694377116e-06,
+      "loss": 0.6644,
+      "step": 1398
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68207997673968e-06,
+      "loss": 0.6432,
+      "step": 1399
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.668852546306409e-06,
+      "loss": 0.6769,
+      "step": 1400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.655629427542636e-06,
+      "loss": 0.7033,
+      "step": 1401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642410644905726e-06,
+      "loss": 0.6687,
+      "step": 1402
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.629196222845027e-06,
+      "loss": 0.6185,
+      "step": 1403
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615986185801807e-06,
+      "loss": 0.681,
+      "step": 1404
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.602780558209234e-06,
+      "loss": 0.6537,
+      "step": 1405
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5895793644923195e-06,
+      "loss": 0.6869,
+      "step": 1406
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.576382629067877e-06,
+      "loss": 0.6529,
+      "step": 1407
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.563190376344467e-06,
+      "loss": 0.6618,
+      "step": 1408
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.550002630722366e-06,
+      "loss": 0.7011,
+      "step": 1409
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.536819416593504e-06,
+      "loss": 0.67,
+      "step": 1410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523640758341441e-06,
+      "loss": 0.6752,
+      "step": 1411
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5104666803413015e-06,
+      "loss": 0.7002,
+      "step": 1412
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497297206959746e-06,
+      "loss": 0.6343,
+      "step": 1413
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484132362554915e-06,
+      "loss": 0.6644,
+      "step": 1414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.470972171476389e-06,
+      "loss": 0.6438,
+      "step": 1415
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4578166580651335e-06,
+      "loss": 0.6983,
+      "step": 1416
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.444665846653471e-06,
+      "loss": 0.6679,
+      "step": 1417
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.431519761565027e-06,
+      "loss": 0.7115,
+      "step": 1418
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4183784271146796e-06,
+      "loss": 0.6758,
+      "step": 1419
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.405241867608526e-06,
+      "loss": 0.7189,
+      "step": 1420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.392110107343833e-06,
+      "loss": 0.662,
+      "step": 1421
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.378983170608982e-06,
+      "loss": 0.6309,
+      "step": 1422
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.365861081683434e-06,
+      "loss": 0.7172,
+      "step": 1423
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.352743864837695e-06,
+      "loss": 0.6679,
+      "step": 1424
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.33963154433325e-06,
+      "loss": 0.6658,
+      "step": 1425
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3265241444225275e-06,
+      "loss": 0.7011,
+      "step": 1426
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.313421689348863e-06,
+      "loss": 0.6687,
+      "step": 1427
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3003242033464314e-06,
+      "loss": 0.6753,
+      "step": 1428
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2872317106402344e-06,
+      "loss": 0.6817,
+      "step": 1429
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.274144235446024e-06,
+      "loss": 0.652,
+      "step": 1430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2610618019702776e-06,
+      "loss": 0.676,
+      "step": 1431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.24798443441015e-06,
+      "loss": 0.6847,
+      "step": 1432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.234912156953424e-06,
+      "loss": 0.6231,
+      "step": 1433
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.221844993778464e-06,
+      "loss": 0.7355,
+      "step": 1434
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.208782969054178e-06,
+      "loss": 0.6294,
+      "step": 1435
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1957261069399745e-06,
+      "loss": 0.6823,
+      "step": 1436
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.6155,
+      "step": 1437
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.16962796713163e-06,
+      "loss": 0.639,
+      "step": 1438
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.156586737708383e-06,
+      "loss": 0.6155,
+      "step": 1439
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143550767436894e-06,
+      "loss": 0.6722,
+      "step": 1440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.130520080428385e-06,
+      "loss": 0.6292,
+      "step": 1441
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.117494700784292e-06,
+      "loss": 0.6025,
+      "step": 1442
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1044746525962446e-06,
+      "loss": 0.6872,
+      "step": 1443
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0914599599460095e-06,
+      "loss": 0.6725,
+      "step": 1444
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.078450646905446e-06,
+      "loss": 0.6843,
+      "step": 1445
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065446737536455e-06,
+      "loss": 0.624,
+      "step": 1446
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.052448255890958e-06,
+      "loss": 0.6521,
+      "step": 1447
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.039455226010832e-06,
+      "loss": 0.6485,
+      "step": 1448
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026467671927863e-06,
+      "loss": 0.6616,
+      "step": 1449
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.01348561766372e-06,
+      "loss": 0.6578,
+      "step": 1450
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0005090872298955e-06,
+      "loss": 0.6796,
+      "step": 1451
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9875381046276605e-06,
+      "loss": 0.6515,
+      "step": 1452
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.974572693848027e-06,
+      "loss": 0.6707,
+      "step": 1453
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.961612878871705e-06,
+      "loss": 0.6543,
+      "step": 1454
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.948658683669056e-06,
+      "loss": 0.6628,
+      "step": 1455
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.93571013220004e-06,
+      "loss": 0.6687,
+      "step": 1456
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.922767248414184e-06,
+      "loss": 0.6171,
+      "step": 1457
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.6367,
+      "step": 1458
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.896898579637588e-06,
+      "loss": 0.5822,
+      "step": 1459
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.883972842493306e-06,
+      "loss": 0.6607,
+      "step": 1460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871052868725012e-06,
+      "loss": 0.6439,
+      "step": 1461
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8581386822293765e-06,
+      "loss": 0.6924,
+      "step": 1462
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.845230306892361e-06,
+      "loss": 0.6888,
+      "step": 1463
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.832327766589177e-06,
+      "loss": 0.6456,
+      "step": 1464
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.819431085184251e-06,
+      "loss": 0.7035,
+      "step": 1465
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806540286531173e-06,
+      "loss": 0.7024,
+      "step": 1466
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.793655394472645e-06,
+      "loss": 0.5852,
+      "step": 1467
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.780776432840449e-06,
+      "loss": 0.6993,
+      "step": 1468
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.767903425455402e-06,
+      "loss": 0.6852,
+      "step": 1469
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755036396127297e-06,
+      "loss": 0.6471,
+      "step": 1470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.742175368654876e-06,
+      "loss": 0.6877,
+      "step": 1471
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.729320366825785e-06,
+      "loss": 0.6705,
+      "step": 1472
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.716471414416519e-06,
+      "loss": 0.6751,
+      "step": 1473
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7036285351923836e-06,
+      "loss": 0.7104,
+      "step": 1474
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6907917529074595e-06,
+      "loss": 0.7003,
+      "step": 1475
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6779610913045344e-06,
+      "loss": 0.6549,
+      "step": 1476
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6651365741150955e-06,
+      "loss": 0.6621,
+      "step": 1477
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.652318225059248e-06,
+      "loss": 0.6773,
+      "step": 1478
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.639506067845698e-06,
+      "loss": 0.6642,
+      "step": 1479
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6267001261717015e-06,
+      "loss": 0.6728,
+      "step": 1480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.613900423723008e-06,
+      "loss": 0.6389,
+      "step": 1481
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.601106984173835e-06,
+      "loss": 0.6401,
+      "step": 1482
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588319831186811e-06,
+      "loss": 0.6821,
+      "step": 1483
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.575538988412944e-06,
+      "loss": 0.6773,
+      "step": 1484
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.562764479491566e-06,
+      "loss": 0.6161,
+      "step": 1485
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.549996328050296e-06,
+      "loss": 0.6747,
+      "step": 1486
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537234557704988e-06,
+      "loss": 0.6632,
+      "step": 1487
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524479192059699e-06,
+      "loss": 0.6507,
+      "step": 1488
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.511730254706643e-06,
+      "loss": 0.6449,
+      "step": 1489
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.498987769226138e-06,
+      "loss": 0.5859,
+      "step": 1490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.486251759186573e-06,
+      "loss": 0.7432,
+      "step": 1491
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.473522248144359e-06,
+      "loss": 0.6079,
+      "step": 1492
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.460799259643884e-06,
+      "loss": 0.6941,
+      "step": 1493
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4480828172174714e-06,
+      "loss": 0.6742,
+      "step": 1494
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.435372944385343e-06,
+      "loss": 0.7017,
+      "step": 1495
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4226696646555675e-06,
+      "loss": 0.6552,
+      "step": 1496
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.409973001524012e-06,
+      "loss": 0.6727,
+      "step": 1497
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.39728297847432e-06,
+      "loss": 0.6823,
+      "step": 1498
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384599618977836e-06,
+      "loss": 0.6572,
+      "step": 1499
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3719229464935915e-06,
+      "loss": 0.6608,
+      "step": 1500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359252984468247e-06,
+      "loss": 0.6497,
+      "step": 1501
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 0.6635,
+      "step": 1502
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.333933285518797e-06,
+      "loss": 0.6335,
+      "step": 1503
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.321283595425784e-06,
+      "loss": 0.6658,
+      "step": 1504
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3086407094537575e-06,
+      "loss": 0.6317,
+      "step": 1505
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.29600465098689e-06,
+      "loss": 0.654,
+      "step": 1506
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.283375443396726e-06,
+      "loss": 0.635,
+      "step": 1507
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270753110042131e-06,
+      "loss": 0.7114,
+      "step": 1508
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258137674269262e-06,
+      "loss": 0.7165,
+      "step": 1509
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2455291594115184e-06,
+      "loss": 0.6271,
+      "step": 1510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.232927588789492e-06,
+      "loss": 0.7492,
+      "step": 1511
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.220332985710936e-06,
+      "loss": 0.6483,
+      "step": 1512
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.207745373470717e-06,
+      "loss": 0.6473,
+      "step": 1513
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19516477535077e-06,
+      "loss": 0.6583,
+      "step": 1514
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.182591214620057e-06,
+      "loss": 0.6649,
+      "step": 1515
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.170024714534526e-06,
+      "loss": 0.7031,
+      "step": 1516
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.157465298337057e-06,
+      "loss": 0.6662,
+      "step": 1517
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.144912989257441e-06,
+      "loss": 0.6135,
+      "step": 1518
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13236781051231e-06,
+      "loss": 0.6201,
+      "step": 1519
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.119829785305118e-06,
+      "loss": 0.6264,
+      "step": 1520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.107298936826086e-06,
+      "loss": 0.6526,
+      "step": 1521
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094775288252157e-06,
+      "loss": 0.6977,
+      "step": 1522
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.082258862746961e-06,
+      "loss": 0.6237,
+      "step": 1523
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.069749683460765e-06,
+      "loss": 0.6519,
+      "step": 1524
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.057247773530437e-06,
+      "loss": 0.7131,
+      "step": 1525
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.044753156079395e-06,
+      "loss": 0.6559,
+      "step": 1526
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.032265854217574e-06,
+      "loss": 0.6806,
+      "step": 1527
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.019785891041381e-06,
+      "loss": 0.645,
+      "step": 1528
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007313289633634e-06,
+      "loss": 0.6892,
+      "step": 1529
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.994848073063552e-06,
+      "loss": 0.6915,
+      "step": 1530
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.982390264386684e-06,
+      "loss": 0.6705,
+      "step": 1531
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9699398866448846e-06,
+      "loss": 0.6372,
+      "step": 1532
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.957496962866263e-06,
+      "loss": 0.7057,
+      "step": 1533
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.94506151606514e-06,
+      "loss": 0.6873,
+      "step": 1534
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 0.6955,
+      "step": 1535
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9202131453834664e-06,
+      "loss": 0.674,
+      "step": 1536
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.907800267462247e-06,
+      "loss": 0.7214,
+      "step": 1537
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.895394958437086e-06,
+      "loss": 0.7414,
+      "step": 1538
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8829972412527325e-06,
+      "loss": 0.6578,
+      "step": 1539
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.870607138839899e-06,
+      "loss": 0.6442,
+      "step": 1540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.858224674115203e-06,
+      "loss": 0.6009,
+      "step": 1541
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.665,
+      "step": 1542
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.833482749326036e-06,
+      "loss": 0.6421,
+      "step": 1543
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8211233350240105e-06,
+      "loss": 0.6731,
+      "step": 1544
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.808771649934924e-06,
+      "loss": 0.6021,
+      "step": 1545
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796427716904347e-06,
+      "loss": 0.6595,
+      "step": 1546
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784091558763501e-06,
+      "loss": 0.6821,
+      "step": 1547
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7717631983292375e-06,
+      "loss": 0.6318,
+      "step": 1548
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759442658403985e-06,
+      "loss": 0.6437,
+      "step": 1549
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747129961775697e-06,
+      "loss": 0.6393,
+      "step": 1550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.73482513121783e-06,
+      "loss": 0.6554,
+      "step": 1551
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.722528189489294e-06,
+      "loss": 0.5667,
+      "step": 1552
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.710239159334393e-06,
+      "loss": 0.6372,
+      "step": 1553
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6979580634828125e-06,
+      "loss": 0.6911,
+      "step": 1554
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.685684924649557e-06,
+      "loss": 0.6335,
+      "step": 1555
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.673419765534915e-06,
+      "loss": 0.6479,
+      "step": 1556
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.6667,
+      "step": 1557
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648913477188793e-06,
+      "loss": 0.6158,
+      "step": 1558
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.636672393283918e-06,
+      "loss": 0.716,
+      "step": 1559
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6244393797507944e-06,
+      "loss": 0.6825,
+      "step": 1560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.612214459215497e-06,
+      "loss": 0.6768,
+      "step": 1561
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.599997654289129e-06,
+      "loss": 0.6819,
+      "step": 1562
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.587788987567785e-06,
+      "loss": 0.6705,
+      "step": 1563
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.575588481632508e-06,
+      "loss": 0.6656,
+      "step": 1564
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5633961590492415e-06,
+      "loss": 0.6374,
+      "step": 1565
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.551212042368792e-06,
+      "loss": 0.6546,
+      "step": 1566
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.539036154126799e-06,
+      "loss": 0.621,
+      "step": 1567
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.526868516843673e-06,
+      "loss": 0.6793,
+      "step": 1568
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514709153024571e-06,
+      "loss": 0.6526,
+      "step": 1569
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502558085159344e-06,
+      "loss": 0.6861,
+      "step": 1570
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490415335722492e-06,
+      "loss": 0.5963,
+      "step": 1571
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.478280927173145e-06,
+      "loss": 0.701,
+      "step": 1572
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466154881954988e-06,
+      "loss": 0.7078,
+      "step": 1573
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.45403722249625e-06,
+      "loss": 0.6305,
+      "step": 1574
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.441927971209644e-06,
+      "loss": 0.6754,
+      "step": 1575
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.429827150492339e-06,
+      "loss": 0.6391,
+      "step": 1576
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.417734782725896e-06,
+      "loss": 0.6544,
+      "step": 1577
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.405650890276255e-06,
+      "loss": 0.6492,
+      "step": 1578
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.393575495493679e-06,
+      "loss": 0.5966,
+      "step": 1579
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381508620712703e-06,
+      "loss": 0.6439,
+      "step": 1580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369450288252113e-06,
+      "loss": 0.6757,
+      "step": 1581
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.357400520414898e-06,
+      "loss": 0.6777,
+      "step": 1582
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.345359339488194e-06,
+      "loss": 0.6858,
+      "step": 1583
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333326767743263e-06,
+      "loss": 0.6917,
+      "step": 1584
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.321302827435445e-06,
+      "loss": 0.7541,
+      "step": 1585
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.309287540804116e-06,
+      "loss": 0.6046,
+      "step": 1586
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.297280930072632e-06,
+      "loss": 0.7561,
+      "step": 1587
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285283017448325e-06,
+      "loss": 0.6355,
+      "step": 1588
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2732938251224165e-06,
+      "loss": 0.456,
+      "step": 1589
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2613133752700145e-06,
+      "loss": 0.4502,
+      "step": 1590
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.249341690050051e-06,
+      "loss": 0.4049,
+      "step": 1591
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.237378791605249e-06,
+      "loss": 0.424,
+      "step": 1592
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.225424702062082e-06,
+      "loss": 0.4494,
+      "step": 1593
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.21347944353072e-06,
+      "loss": 0.4324,
+      "step": 1594
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.201543038105012e-06,
+      "loss": 0.4091,
+      "step": 1595
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1896155078624225e-06,
+      "loss": 0.4082,
+      "step": 1596
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1776968748640065e-06,
+      "loss": 0.4218,
+      "step": 1597
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.165787161154361e-06,
+      "loss": 0.3819,
+      "step": 1598
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.153886388761586e-06,
+      "loss": 0.3985,
+      "step": 1599
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.14199457969725e-06,
+      "loss": 0.4318,
+      "step": 1600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.130111755956327e-06,
+      "loss": 0.4111,
+      "step": 1601
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.11823793951719e-06,
+      "loss": 0.4255,
+      "step": 1602
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.106373152341535e-06,
+      "loss": 0.3979,
+      "step": 1603
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094517416374371e-06,
+      "loss": 0.4018,
+      "step": 1604
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.082670753543961e-06,
+      "loss": 0.4164,
+      "step": 1605
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.070833185761792e-06,
+      "loss": 0.3996,
+      "step": 1606
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0590047349225135e-06,
+      "loss": 0.421,
+      "step": 1607
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0471854229039286e-06,
+      "loss": 0.3969,
+      "step": 1608
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.035375271566933e-06,
+      "loss": 0.3819,
+      "step": 1609
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.023574302755471e-06,
+      "loss": 0.3986,
+      "step": 1610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011782538296512e-06,
+      "loss": 0.4277,
+      "step": 1611
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.454,
+      "step": 1612
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.988226709658812e-06,
+      "loss": 0.4285,
+      "step": 1613
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.976462689048718e-06,
+      "loss": 0.4019,
+      "step": 1614
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964707959928345e-06,
+      "loss": 0.4116,
+      "step": 1615
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.952962544039143e-06,
+      "loss": 0.4269,
+      "step": 1616
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.941226463105322e-06,
+      "loss": 0.4302,
+      "step": 1617
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.929499738833828e-06,
+      "loss": 0.387,
+      "step": 1618
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.917782392914311e-06,
+      "loss": 0.446,
+      "step": 1619
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.906074447019068e-06,
+      "loss": 0.4256,
+      "step": 1620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.894375922803012e-06,
+      "loss": 0.3853,
+      "step": 1621
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.882686841903627e-06,
+      "loss": 0.4203,
+      "step": 1622
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.871007225940939e-06,
+      "loss": 0.4418,
+      "step": 1623
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.859337096517457e-06,
+      "loss": 0.4321,
+      "step": 1624
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.847676475218143e-06,
+      "loss": 0.4047,
+      "step": 1625
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836025383610382e-06,
+      "loss": 0.4354,
+      "step": 1626
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.824383843243929e-06,
+      "loss": 0.4242,
+      "step": 1627
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812751875650875e-06,
+      "loss": 0.4104,
+      "step": 1628
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801129502345605e-06,
+      "loss": 0.4002,
+      "step": 1629
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7895167448247494e-06,
+      "loss": 0.3929,
+      "step": 1630
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.777913624567167e-06,
+      "loss": 0.4243,
+      "step": 1631
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.766320163033882e-06,
+      "loss": 0.385,
+      "step": 1632
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754736381668057e-06,
+      "loss": 0.4747,
+      "step": 1633
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743162301894952e-06,
+      "loss": 0.4349,
+      "step": 1634
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7315979451218866e-06,
+      "loss": 0.4118,
+      "step": 1635
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.720043332738182e-06,
+      "loss": 0.4126,
+      "step": 1636
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7084984861151504e-06,
+      "loss": 0.4032,
+      "step": 1637
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.696963426606041e-06,
+      "loss": 0.4401,
+      "step": 1638
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.68543817554599e-06,
+      "loss": 0.4242,
+      "step": 1639
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.673922754252001e-06,
+      "loss": 0.4331,
+      "step": 1640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6624171840229005e-06,
+      "loss": 0.3979,
+      "step": 1641
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6509214861392785e-06,
+      "loss": 0.4101,
+      "step": 1642
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.639435681863479e-06,
+      "loss": 0.3774,
+      "step": 1643
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6279597924395434e-06,
+      "loss": 0.4004,
+      "step": 1644
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.616493839093179e-06,
+      "loss": 0.4223,
+      "step": 1645
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.605037843031702e-06,
+      "loss": 0.4188,
+      "step": 1646
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.4282,
+      "step": 1647
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.582155807500601e-06,
+      "loss": 0.4022,
+      "step": 1648
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5707298103533835e-06,
+      "loss": 0.3792,
+      "step": 1649
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559313855135795e-06,
+      "loss": 0.4064,
+      "step": 1650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.547907962962685e-06,
+      "loss": 0.4114,
+      "step": 1651
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5365121549302916e-06,
+      "loss": 0.3875,
+      "step": 1652
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.525126452116201e-06,
+      "loss": 0.3466,
+      "step": 1653
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513750875579303e-06,
+      "loss": 0.3995,
+      "step": 1654
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.502385446359763e-06,
+      "loss": 0.4301,
+      "step": 1655
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.491030185478976e-06,
+      "loss": 0.3997,
+      "step": 1656
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.479685113939533e-06,
+      "loss": 0.4062,
+      "step": 1657
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.468350252725178e-06,
+      "loss": 0.4065,
+      "step": 1658
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457025622800771e-06,
+      "loss": 0.3936,
+      "step": 1659
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.445711245112241e-06,
+      "loss": 0.4158,
+      "step": 1660
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4344071405865656e-06,
+      "loss": 0.4569,
+      "step": 1661
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423113330131708e-06,
+      "loss": 0.4131,
+      "step": 1662
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.411829834636603e-06,
+      "loss": 0.4095,
+      "step": 1663
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.400556674971104e-06,
+      "loss": 0.4438,
+      "step": 1664
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.389293871985949e-06,
+      "loss": 0.4251,
+      "step": 1665
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.378041446512711e-06,
+      "loss": 0.4351,
+      "step": 1666
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3667994193637794e-06,
+      "loss": 0.3707,
+      "step": 1667
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.355567811332311e-06,
+      "loss": 0.4218,
+      "step": 1668
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.344346643192179e-06,
+      "loss": 0.3521,
+      "step": 1669
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.333135935697961e-06,
+      "loss": 0.4216,
+      "step": 1670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.321935709584884e-06,
+      "loss": 0.3853,
+      "step": 1671
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.310745985568779e-06,
+      "loss": 0.3789,
+      "step": 1672
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.299566784346063e-06,
+      "loss": 0.3824,
+      "step": 1673
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2883981265936884e-06,
+      "loss": 0.4289,
+      "step": 1674
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2772400329691055e-06,
+      "loss": 0.3549,
+      "step": 1675
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266092524110219e-06,
+      "loss": 0.4609,
+      "step": 1676
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.254955620635371e-06,
+      "loss": 0.4205,
+      "step": 1677
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2438293431432665e-06,
+      "loss": 0.3819,
+      "step": 1678
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.232713712212977e-06,
+      "loss": 0.3947,
+      "step": 1679
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221608748403872e-06,
+      "loss": 0.3876,
+      "step": 1680
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.210514472255594e-06,
+      "loss": 0.4124,
+      "step": 1681
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.19943090428802e-06,
+      "loss": 0.4141,
+      "step": 1682
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.188358065001215e-06,
+      "loss": 0.4022,
+      "step": 1683
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.177295974875401e-06,
+      "loss": 0.3667,
+      "step": 1684
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.166244654370922e-06,
+      "loss": 0.3967,
+      "step": 1685
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.155204123928205e-06,
+      "loss": 0.4314,
+      "step": 1686
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144174403967714e-06,
+      "loss": 0.4123,
+      "step": 1687
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1331555148899215e-06,
+      "loss": 0.3897,
+      "step": 1688
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.4391,
+      "step": 1689
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.11115031088412e-06,
+      "loss": 0.3589,
+      "step": 1690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.10016403665674e-06,
+      "loss": 0.4371,
+      "step": 1691
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0891886747132356e-06,
+      "loss": 0.4759,
+      "step": 1692
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.078224245353545e-06,
+      "loss": 0.4241,
+      "step": 1693
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.067270768857375e-06,
+      "loss": 0.4503,
+      "step": 1694
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.056328265484184e-06,
+      "loss": 0.4124,
+      "step": 1695
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.045396755473121e-06,
+      "loss": 0.3926,
+      "step": 1696
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034476259043012e-06,
+      "loss": 0.3785,
+      "step": 1697
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.023566796392313e-06,
+      "loss": 0.4244,
+      "step": 1698
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0126683876990626e-06,
+      "loss": 0.4294,
+      "step": 1699
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.001781053120863e-06,
+      "loss": 0.4163,
+      "step": 1700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.990904812794835e-06,
+      "loss": 0.3934,
+      "step": 1701
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9800396868375675e-06,
+      "loss": 0.3938,
+      "step": 1702
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9691856953451044e-06,
+      "loss": 0.3929,
+      "step": 1703
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.958342858392893e-06,
+      "loss": 0.4191,
+      "step": 1704
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9475111960357485e-06,
+      "loss": 0.443,
+      "step": 1705
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93669072830781e-06,
+      "loss": 0.4255,
+      "step": 1706
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925881475222528e-06,
+      "loss": 0.382,
+      "step": 1707
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915083456772591e-06,
+      "loss": 0.3895,
+      "step": 1708
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90429669292992e-06,
+      "loss": 0.3494,
+      "step": 1709
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.893521203645618e-06,
+      "loss": 0.4241,
+      "step": 1710
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882757008849936e-06,
+      "loss": 0.4436,
+      "step": 1711
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.872004128452233e-06,
+      "loss": 0.399,
+      "step": 1712
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8612625823409365e-06,
+      "loss": 0.3655,
+      "step": 1713
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8505323903835134e-06,
+      "loss": 0.4032,
+      "step": 1714
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8398135724264326e-06,
+      "loss": 0.4238,
+      "step": 1715
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.829106148295127e-06,
+      "loss": 0.3965,
+      "step": 1716
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.818410137793947e-06,
+      "loss": 0.3858,
+      "step": 1717
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8077255607061425e-06,
+      "loss": 0.3754,
+      "step": 1718
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7970524367938146e-06,
+      "loss": 0.437,
+      "step": 1719
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.786390785797873e-06,
+      "loss": 0.3756,
+      "step": 1720
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7757406274380094e-06,
+      "loss": 0.4348,
+      "step": 1721
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.3985,
+      "step": 1722
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.754474867398986e-06,
+      "loss": 0.3501,
+      "step": 1723
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.743859305052785e-06,
+      "loss": 0.3842,
+      "step": 1724
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.733255314008516e-06,
+      "loss": 0.4598,
+      "step": 1725
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.722662913879218e-06,
+      "loss": 0.4603,
+      "step": 1726
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7120821242565086e-06,
+      "loss": 0.3999,
+      "step": 1727
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.701512964710513e-06,
+      "loss": 0.4109,
+      "step": 1728
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6909554547898585e-06,
+      "loss": 0.4559,
+      "step": 1729
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6804096140216217e-06,
+      "loss": 0.4203,
+      "step": 1730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6698754619112974e-06,
+      "loss": 0.4554,
+      "step": 1731
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.659353017942754e-06,
+      "loss": 0.381,
+      "step": 1732
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6488423015782128e-06,
+      "loss": 0.393,
+      "step": 1733
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6383433322582028e-06,
+      "loss": 0.4161,
+      "step": 1734
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.627856129401519e-06,
+      "loss": 0.373,
+      "step": 1735
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6173807124051996e-06,
+      "loss": 0.3929,
+      "step": 1736
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6069171006444882e-06,
+      "loss": 0.3889,
+      "step": 1737
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.596465313472778e-06,
+      "loss": 0.3984,
+      "step": 1738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5860253702216062e-06,
+      "loss": 0.4151,
+      "step": 1739
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.575597290200599e-06,
+      "loss": 0.4046,
+      "step": 1740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5651810926974382e-06,
+      "loss": 0.4333,
+      "step": 1741
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5547767969778355e-06,
+      "loss": 0.4239,
+      "step": 1742
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.544384422285477e-06,
+      "loss": 0.4024,
+      "step": 1743
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.534003987842005e-06,
+      "loss": 0.4348,
+      "step": 1744
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5236355128469814e-06,
+      "loss": 0.3993,
+      "step": 1745
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.513279016477844e-06,
+      "loss": 0.3902,
+      "step": 1746
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5029345178898756e-06,
+      "loss": 0.3927,
+      "step": 1747
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.492602036216174e-06,
+      "loss": 0.4081,
+      "step": 1748
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4822815905675955e-06,
+      "loss": 0.4231,
+      "step": 1749
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4719732000327524e-06,
+      "loss": 0.4253,
+      "step": 1750
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4616768836779448e-06,
+      "loss": 0.4134,
+      "step": 1751
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.4084,
+      "step": 1752
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.441120549661976e-06,
+      "loss": 0.4064,
+      "step": 1753
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4308605700216315e-06,
+      "loss": 0.3518,
+      "step": 1754
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4206127406028744e-06,
+      "loss": 0.4034,
+      "step": 1755
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4103770803600033e-06,
+      "loss": 0.4087,
+      "step": 1756
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4001536082248075e-06,
+      "loss": 0.3899,
+      "step": 1757
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3899423431065215e-06,
+      "loss": 0.4046,
+      "step": 1758
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.379743303891815e-06,
+      "loss": 0.4106,
+      "step": 1759
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3695565094447423e-06,
+      "loss": 0.379,
+      "step": 1760
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3593819786067005e-06,
+      "loss": 0.3873,
+      "step": 1761
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3492197301964145e-06,
+      "loss": 0.4162,
+      "step": 1762
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3390697830098884e-06,
+      "loss": 0.3608,
+      "step": 1763
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3289321558203767e-06,
+      "loss": 0.3906,
+      "step": 1764
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.318806867378338e-06,
+      "loss": 0.3638,
+      "step": 1765
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.4554,
+      "step": 1766
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2985933816244065e-06,
+      "loss": 0.4003,
+      "step": 1767
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.288505221699194e-06,
+      "loss": 0.396,
+      "step": 1768
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2784294752947545e-06,
+      "loss": 0.3955,
+      "step": 1769
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.268366161047096e-06,
+      "loss": 0.4253,
+      "step": 1770
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.258315297569241e-06,
+      "loss": 0.4001,
+      "step": 1771
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.248276903451171e-06,
+      "loss": 0.4475,
+      "step": 1772
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2382509972598087e-06,
+      "loss": 0.4006,
+      "step": 1773
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2282375975389825e-06,
+      "loss": 0.3713,
+      "step": 1774
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.218236722809387e-06,
+      "loss": 0.3918,
+      "step": 1775
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2082483915685526e-06,
+      "loss": 0.3678,
+      "step": 1776
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1982726222908046e-06,
+      "loss": 0.3679,
+      "step": 1777
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1883094334272424e-06,
+      "loss": 0.3911,
+      "step": 1778
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1783588434056845e-06,
+      "loss": 0.3962,
+      "step": 1779
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1684208706306572e-06,
+      "loss": 0.3758,
+      "step": 1780
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1584955334833434e-06,
+      "loss": 0.3978,
+      "step": 1781
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1485828503215588e-06,
+      "loss": 0.404,
+      "step": 1782
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.138682839479713e-06,
+      "loss": 0.4401,
+      "step": 1783
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1287955192687834e-06,
+      "loss": 0.4032,
+      "step": 1784
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1189209079762606e-06,
+      "loss": 0.4437,
+      "step": 1785
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.109059023866142e-06,
+      "loss": 0.4275,
+      "step": 1786
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.099209885178882e-06,
+      "loss": 0.3973,
+      "step": 1787
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.089373510131354e-06,
+      "loss": 0.3918,
+      "step": 1788
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.079549916916833e-06,
+      "loss": 0.3891,
+      "step": 1789
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0697391237049514e-06,
+      "loss": 0.4166,
+      "step": 1790
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0599411486416587e-06,
+      "loss": 0.4259,
+      "step": 1791
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0501560098492056e-06,
+      "loss": 0.3664,
+      "step": 1792
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.040383725426098e-06,
+      "loss": 0.3898,
+      "step": 1793
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0306243134470668e-06,
+      "loss": 0.418,
+      "step": 1794
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0208777919630284e-06,
+      "loss": 0.4308,
+      "step": 1795
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0111441790010687e-06,
+      "loss": 0.3846,
+      "step": 1796
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.001423492564384e-06,
+      "loss": 0.3903,
+      "step": 1797
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9917157506322727e-06,
+      "loss": 0.4282,
+      "step": 1798
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9820209711600858e-06,
+      "loss": 0.4056,
+      "step": 1799
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.972339172079204e-06,
+      "loss": 0.3793,
+      "step": 1800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9626703712969962e-06,
+      "loss": 0.3976,
+      "step": 1801
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9530145866967897e-06,
+      "loss": 0.4132,
+      "step": 1802
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9433718361378326e-06,
+      "loss": 0.4034,
+      "step": 1803
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9337421374552733e-06,
+      "loss": 0.3926,
+      "step": 1804
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.924125508460117e-06,
+      "loss": 0.4423,
+      "step": 1805
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9145219669391944e-06,
+      "loss": 0.4379,
+      "step": 1806
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9049315306551306e-06,
+      "loss": 0.3887,
+      "step": 1807
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8953542173463133e-06,
+      "loss": 0.3951,
+      "step": 1808
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.885790044726853e-06,
+      "loss": 0.3944,
+      "step": 1809
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.876239030486554e-06,
+      "loss": 0.4,
+      "step": 1810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8667011922908907e-06,
+      "loss": 0.3713,
+      "step": 1811
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8571765477809645e-06,
+      "loss": 0.3924,
+      "step": 1812
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8476651145734703e-06,
+      "loss": 0.437,
+      "step": 1813
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8381669102606746e-06,
+      "loss": 0.4069,
+      "step": 1814
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8286819524103657e-06,
+      "loss": 0.3853,
+      "step": 1815
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.819210258565841e-06,
+      "loss": 0.3917,
+      "step": 1816
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.809751846245856e-06,
+      "loss": 0.4045,
+      "step": 1817
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.80030673294461e-06,
+      "loss": 0.3954,
+      "step": 1818
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7908749361316966e-06,
+      "loss": 0.4044,
+      "step": 1819
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.78145647325209e-06,
+      "loss": 0.3926,
+      "step": 1820
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7720513617260857e-06,
+      "loss": 0.3807,
+      "step": 1821
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7626596189492983e-06,
+      "loss": 0.4193,
+      "step": 1822
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7532812622926143e-06,
+      "loss": 0.4325,
+      "step": 1823
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7439163091021525e-06,
+      "loss": 0.4471,
+      "step": 1824
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7345647766992487e-06,
+      "loss": 0.4457,
+      "step": 1825
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7252266823804184e-06,
+      "loss": 0.4135,
+      "step": 1826
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.71590204341731e-06,
+      "loss": 0.4427,
+      "step": 1827
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.706590877056696e-06,
+      "loss": 0.4287,
+      "step": 1828
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6972932005204267e-06,
+      "loss": 0.3845,
+      "step": 1829
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.688009031005403e-06,
+      "loss": 0.4344,
+      "step": 1830
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.678738385683538e-06,
+      "loss": 0.4133,
+      "step": 1831
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 0.3441,
+      "step": 1832
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6602377361818576e-06,
+      "loss": 0.394,
+      "step": 1833
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6510077662206768e-06,
+      "loss": 0.4133,
+      "step": 1834
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.641791388889864e-06,
+      "loss": 0.4094,
+      "step": 1835
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6325886212359496e-06,
+      "loss": 0.4195,
+      "step": 1836
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.623399480280292e-06,
+      "loss": 0.3734,
+      "step": 1837
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6142239830190453e-06,
+      "loss": 0.4343,
+      "step": 1838
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6050621464231243e-06,
+      "loss": 0.4012,
+      "step": 1839
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5959139874381778e-06,
+      "loss": 0.3691,
+      "step": 1840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5867795229845627e-06,
+      "loss": 0.3739,
+      "step": 1841
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5776587699573007e-06,
+      "loss": 0.4074,
+      "step": 1842
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 0.3987,
+      "step": 1843
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.559458465635104e-06,
+      "loss": 0.3508,
+      "step": 1844
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.550378948003287e-06,
+      "loss": 0.3951,
+      "step": 1845
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5413132091240057e-06,
+      "loss": 0.3837,
+      "step": 1846
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5322612657651637e-06,
+      "loss": 0.4238,
+      "step": 1847
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.523223134669157e-06,
+      "loss": 0.3921,
+      "step": 1848
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.514198832552832e-06,
+      "loss": 0.3947,
+      "step": 1849
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5051883761074613e-06,
+      "loss": 0.4449,
+      "step": 1850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496191781998698e-06,
+      "loss": 0.4149,
+      "step": 1851
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.487209066866565e-06,
+      "loss": 0.3759,
+      "step": 1852
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478240247325415e-06,
+      "loss": 0.4015,
+      "step": 1853
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.469285339963892e-06,
+      "loss": 0.3944,
+      "step": 1854
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.460344361344913e-06,
+      "loss": 0.3938,
+      "step": 1855
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4514173280056363e-06,
+      "loss": 0.4186,
+      "step": 1856
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 0.3935,
+      "step": 1857
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.433605163185798e-06,
+      "loss": 0.4298,
+      "step": 1858
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424720064650459e-06,
+      "loss": 0.3727,
+      "step": 1859
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4158489772852035e-06,
+      "loss": 0.4607,
+      "step": 1860
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.406991917497908e-06,
+      "loss": 0.3613,
+      "step": 1861
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.398148901670521e-06,
+      "loss": 0.4099,
+      "step": 1862
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3893199461589943e-06,
+      "loss": 0.3814,
+      "step": 1863
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.380505067293293e-06,
+      "loss": 0.3614,
+      "step": 1864
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.371704281377335e-06,
+      "loss": 0.359,
+      "step": 1865
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3629176046889755e-06,
+      "loss": 0.3945,
+      "step": 1866
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3541450534799783e-06,
+      "loss": 0.4746,
+      "step": 1867
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.345386643975971e-06,
+      "loss": 0.3953,
+      "step": 1868
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3366423923764268e-06,
+      "loss": 0.3958,
+      "step": 1869
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.32791231485464e-06,
+      "loss": 0.421,
+      "step": 1870
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3191964275576806e-06,
+      "loss": 0.4294,
+      "step": 1871
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3104947466063785e-06,
+      "loss": 0.3643,
+      "step": 1872
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.301807288095288e-06,
+      "loss": 0.4525,
+      "step": 1873
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2931340680926485e-06,
+      "loss": 0.362,
+      "step": 1874
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.284475102640371e-06,
+      "loss": 0.4016,
+      "step": 1875
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275830407754006e-06,
+      "loss": 0.4118,
+      "step": 1876
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.267199999422698e-06,
+      "loss": 0.333,
+      "step": 1877
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2585838936091753e-06,
+      "loss": 0.4121,
+      "step": 1878
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.249982106249713e-06,
+      "loss": 0.3896,
+      "step": 1879
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2413946532540953e-06,
+      "loss": 0.4099,
+      "step": 1880
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2328215505056004e-06,
+      "loss": 0.3797,
+      "step": 1881
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.224262813860962e-06,
+      "loss": 0.3702,
+      "step": 1882
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.215718459150348e-06,
+      "loss": 0.4332,
+      "step": 1883
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.207188502177313e-06,
+      "loss": 0.3931,
+      "step": 1884
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.198672958718796e-06,
+      "loss": 0.3692,
+      "step": 1885
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.190171844525063e-06,
+      "loss": 0.3713,
+      "step": 1886
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.4742,
+      "step": 1887
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1732129667995826e-06,
+      "loss": 0.3811,
+      "step": 1888
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1647552346348233e-06,
+      "loss": 0.3863,
+      "step": 1889
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.156311994468774e-06,
+      "loss": 0.4361,
+      "step": 1890
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1478832619179736e-06,
+      "loss": 0.3933,
+      "step": 1891
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1394690525721275e-06,
+      "loss": 0.4625,
+      "step": 1892
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1310693819940842e-06,
+      "loss": 0.4031,
+      "step": 1893
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.122684265719801e-06,
+      "loss": 0.3746,
+      "step": 1894
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1143137192583117e-06,
+      "loss": 0.3905,
+      "step": 1895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1059577580917067e-06,
+      "loss": 0.415,
+      "step": 1896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0976163976750985e-06,
+      "loss": 0.44,
+      "step": 1897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.08928965343659e-06,
+      "loss": 0.4268,
+      "step": 1898
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0809775407772505e-06,
+      "loss": 0.4086,
+      "step": 1899
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.072680075071093e-06,
+      "loss": 0.396,
+      "step": 1900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.064397271665034e-06,
+      "loss": 0.3911,
+      "step": 1901
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0561291458788736e-06,
+      "loss": 0.3979,
+      "step": 1902
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0478757130052662e-06,
+      "loss": 0.4061,
+      "step": 1903
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.039636988309681e-06,
+      "loss": 0.426,
+      "step": 1904
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.031412987030398e-06,
+      "loss": 0.4012,
+      "step": 1905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0232037243784475e-06,
+      "loss": 0.4452,
+      "step": 1906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0150092155376133e-06,
+      "loss": 0.4366,
+      "step": 1907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0068294756643846e-06,
+      "loss": 0.388,
+      "step": 1908
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9986645198879385e-06,
+      "loss": 0.3885,
+      "step": 1909
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9905143633100987e-06,
+      "loss": 0.4084,
+      "step": 1910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.982379021005325e-06,
+      "loss": 0.4338,
+      "step": 1911
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9742585080206754e-06,
+      "loss": 0.3887,
+      "step": 1912
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9661528393757744e-06,
+      "loss": 0.3799,
+      "step": 1913
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.958062030062795e-06,
+      "loss": 0.4152,
+      "step": 1914
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.949986095046431e-06,
+      "loss": 0.3646,
+      "step": 1915
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.941925049263852e-06,
+      "loss": 0.4322,
+      "step": 1916
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9338789076247e-06,
+      "loss": 0.4465,
+      "step": 1917
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9258476850110465e-06,
+      "loss": 0.4057,
+      "step": 1918
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.917831396277372e-06,
+      "loss": 0.4175,
+      "step": 1919
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.4074,
+      "step": 1920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9018436797297234e-06,
+      "loss": 0.388,
+      "step": 1921
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8938722814864863e-06,
+      "loss": 0.3971,
+      "step": 1922
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8859158762646468e-06,
+      "loss": 0.4379,
+      "step": 1923
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8779744787803001e-06,
+      "loss": 0.4126,
+      "step": 1924
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8700481037217843e-06,
+      "loss": 0.423,
+      "step": 1925
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8621367657496504e-06,
+      "loss": 0.4249,
+      "step": 1926
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.854240479496643e-06,
+      "loss": 0.4198,
+      "step": 1927
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8463592595676594e-06,
+      "loss": 0.4195,
+      "step": 1928
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8384931205397305e-06,
+      "loss": 0.3862,
+      "step": 1929
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8306420769619993e-06,
+      "loss": 0.3928,
+      "step": 1930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8228061433556866e-06,
+      "loss": 0.4103,
+      "step": 1931
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8149853342140644e-06,
+      "loss": 0.3839,
+      "step": 1932
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8071796640024341e-06,
+      "loss": 0.4316,
+      "step": 1933
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7993891471580894e-06,
+      "loss": 0.4049,
+      "step": 1934
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7916137980903047e-06,
+      "loss": 0.4316,
+      "step": 1935
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7838536311802912e-06,
+      "loss": 0.4275,
+      "step": 1936
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7761086607811862e-06,
+      "loss": 0.4346,
+      "step": 1937
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7683789012180196e-06,
+      "loss": 0.4065,
+      "step": 1938
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7606643667876867e-06,
+      "loss": 0.4013,
+      "step": 1939
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7529650717589174e-06,
+      "loss": 0.423,
+      "step": 1940
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.74528103037226e-06,
+      "loss": 0.4163,
+      "step": 1941
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 0.3682,
+      "step": 1942
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7299587653463846e-06,
+      "loss": 0.3786,
+      "step": 1943
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.722320570047089e-06,
+      "loss": 0.383,
+      "step": 1944
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7146976850697039e-06,
+      "loss": 0.4042,
+      "step": 1945
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7070901245134463e-06,
+      "loss": 0.4262,
+      "step": 1946
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6994979024491943e-06,
+      "loss": 0.3769,
+      "step": 1947
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6919210329194535e-06,
+      "loss": 0.3525,
+      "step": 1948
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6843595299383365e-06,
+      "loss": 0.3831,
+      "step": 1949
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6768134074915277e-06,
+      "loss": 0.4178,
+      "step": 1950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6692826795362703e-06,
+      "loss": 0.3943,
+      "step": 1951
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6617673600013295e-06,
+      "loss": 0.3603,
+      "step": 1952
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6542674627869738e-06,
+      "loss": 0.385,
+      "step": 1953
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6467830017649467e-06,
+      "loss": 0.3759,
+      "step": 1954
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6393139907784405e-06,
+      "loss": 0.3803,
+      "step": 1955
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6318604436420738e-06,
+      "loss": 0.3499,
+      "step": 1956
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6244223741418574e-06,
+      "loss": 0.4073,
+      "step": 1957
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6169997960351758e-06,
+      "loss": 0.4209,
+      "step": 1958
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.609592723050767e-06,
+      "loss": 0.4013,
+      "step": 1959
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6022011688886853e-06,
+      "loss": 0.3881,
+      "step": 1960
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5948251472202836e-06,
+      "loss": 0.397,
+      "step": 1961
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.4291,
+      "step": 1962
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5801197559062665e-06,
+      "loss": 0.3817,
+      "step": 1963
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5727904134596084e-06,
+      "loss": 0.4242,
+      "step": 1964
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5654766579045033e-06,
+      "loss": 0.381,
+      "step": 1965
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5581785027684038e-06,
+      "loss": 0.3819,
+      "step": 1966
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5508959615499142e-06,
+      "loss": 0.4019,
+      "step": 1967
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5436290477187589e-06,
+      "loss": 0.3984,
+      "step": 1968
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.536377774715757e-06,
+      "loss": 0.4318,
+      "step": 1969
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5291421559527963e-06,
+      "loss": 0.3535,
+      "step": 1970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5219222048128125e-06,
+      "loss": 0.4286,
+      "step": 1971
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5147179346497665e-06,
+      "loss": 0.4192,
+      "step": 1972
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5075293587886065e-06,
+      "loss": 0.3824,
+      "step": 1973
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.500356490525261e-06,
+      "loss": 0.3892,
+      "step": 1974
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4931993431266056e-06,
+      "loss": 0.4094,
+      "step": 1975
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4860579298304311e-06,
+      "loss": 0.4221,
+      "step": 1976
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.478932263845435e-06,
+      "loss": 0.3884,
+      "step": 1977
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4718223583511837e-06,
+      "loss": 0.3724,
+      "step": 1978
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4647282264980988e-06,
+      "loss": 0.4,
+      "step": 1979
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.457649881407417e-06,
+      "loss": 0.3491,
+      "step": 1980
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4505873361711887e-06,
+      "loss": 0.4634,
+      "step": 1981
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.443540603852227e-06,
+      "loss": 0.3558,
+      "step": 1982
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.436509697484111e-06,
+      "loss": 0.4505,
+      "step": 1983
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4294946300711387e-06,
+      "loss": 0.3665,
+      "step": 1984
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4224954145883197e-06,
+      "loss": 0.4084,
+      "step": 1985
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4155120639813392e-06,
+      "loss": 0.4368,
+      "step": 1986
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4085445911665407e-06,
+      "loss": 0.3496,
+      "step": 1987
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4015930090308949e-06,
+      "loss": 0.3781,
+      "step": 1988
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.39465733043199e-06,
+      "loss": 0.3588,
+      "step": 1989
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3877375681979944e-06,
+      "loss": 0.4019,
+      "step": 1990
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.380833735127639e-06,
+      "loss": 0.3747,
+      "step": 1991
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.373945843990192e-06,
+      "loss": 0.3874,
+      "step": 1992
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3670739075254358e-06,
+      "loss": 0.4236,
+      "step": 1993
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3602179384436432e-06,
+      "loss": 0.4288,
+      "step": 1994
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3533779494255483e-06,
+      "loss": 0.3614,
+      "step": 1995
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3465539531223381e-06,
+      "loss": 0.4033,
+      "step": 1996
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.4054,
+      "step": 1997
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.332953989117377e-06,
+      "loss": 0.4126,
+      "step": 1998
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.326178046569997e-06,
+      "loss": 0.3805,
+      "step": 1999
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3194181470461975e-06,
+      "loss": 0.4089,
+      "step": 2000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3126743030490307e-06,
+      "loss": 0.4534,
+      "step": 2001
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3059465270518469e-06,
+      "loss": 0.4027,
+      "step": 2002
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2992348314982818e-06,
+      "loss": 0.3984,
+      "step": 2003
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2925392288022299e-06,
+      "loss": 0.4178,
+      "step": 2004
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2858597313478137e-06,
+      "loss": 0.377,
+      "step": 2005
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2791963514893735e-06,
+      "loss": 0.4496,
+      "step": 2006
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.272549101551438e-06,
+      "loss": 0.4076,
+      "step": 2007
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2659179938287035e-06,
+      "loss": 0.4033,
+      "step": 2008
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.259303040586003e-06,
+      "loss": 0.3969,
+      "step": 2009
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2527042540583e-06,
+      "loss": 0.4038,
+      "step": 2010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2461216464506454e-06,
+      "loss": 0.4085,
+      "step": 2011
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2395552299381742e-06,
+      "loss": 0.422,
+      "step": 2012
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2330050166660712e-06,
+      "loss": 0.3523,
+      "step": 2013
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2264710187495532e-06,
+      "loss": 0.3922,
+      "step": 2014
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2199532482738464e-06,
+      "loss": 0.3982,
+      "step": 2015
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2134517172941563e-06,
+      "loss": 0.3809,
+      "step": 2016
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2069664378356604e-06,
+      "loss": 0.3938,
+      "step": 2017
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2004974218934695e-06,
+      "loss": 0.4131,
+      "step": 2018
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.19404468143262e-06,
+      "loss": 0.3926,
+      "step": 2019
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.187608228388044e-06,
+      "loss": 0.3835,
+      "step": 2020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1811880746645466e-06,
+      "loss": 0.3926,
+      "step": 2021
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1747842321367886e-06,
+      "loss": 0.4242,
+      "step": 2022
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1683967126492557e-06,
+      "loss": 0.3823,
+      "step": 2023
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1620255280162517e-06,
+      "loss": 0.351,
+      "step": 2024
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1556706900218572e-06,
+      "loss": 0.4416,
+      "step": 2025
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.149332210419928e-06,
+      "loss": 0.4535,
+      "step": 2026
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1430101009340576e-06,
+      "loss": 0.3713,
+      "step": 2027
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1367043732575666e-06,
+      "loss": 0.4273,
+      "step": 2028
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1304150390534673e-06,
+      "loss": 0.4013,
+      "step": 2029
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.124142109954459e-06,
+      "loss": 0.4011,
+      "step": 2030
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1178855975628967e-06,
+      "loss": 0.4125,
+      "step": 2031
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1116455134507665e-06,
+      "loss": 0.3937,
+      "step": 2032
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.105421869159673e-06,
+      "loss": 0.3916,
+      "step": 2033
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.099214676200816e-06,
+      "loss": 0.3962,
+      "step": 2034
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0930239460549597e-06,
+      "loss": 0.4434,
+      "step": 2035
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0868496901724245e-06,
+      "loss": 0.4222,
+      "step": 2036
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0806919199730614e-06,
+      "loss": 0.3913,
+      "step": 2037
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0745506468462274e-06,
+      "loss": 0.4201,
+      "step": 2038
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.068425882150762e-06,
+      "loss": 0.4086,
+      "step": 2039
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0623176372149802e-06,
+      "loss": 0.4157,
+      "step": 2040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0562259233366334e-06,
+      "loss": 0.4318,
+      "step": 2041
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0501507517829012e-06,
+      "loss": 0.3635,
+      "step": 2042
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0440921337903699e-06,
+      "loss": 0.4089,
+      "step": 2043
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.038050080565003e-06,
+      "loss": 0.3443,
+      "step": 2044
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0320246032821291e-06,
+      "loss": 0.3723,
+      "step": 2045
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0260157130864178e-06,
+      "loss": 0.4109,
+      "step": 2046
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0200234210918558e-06,
+      "loss": 0.4079,
+      "step": 2047
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0140477383817337e-06,
+      "loss": 0.415,
+      "step": 2048
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0080886760086227e-06,
+      "loss": 0.4123,
+      "step": 2049
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0021462449943497e-06,
+      "loss": 0.3658,
+      "step": 2050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.962204563299826e-07,
+      "loss": 0.3949,
+      "step": 2051
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.4249,
+      "step": 2052
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.844188498613117e-07,
+      "loss": 0.3739,
+      "step": 2053
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.785430538851493e-07,
+      "loss": 0.3915,
+      "step": 2054
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.726839439151447e-07,
+      "loss": 0.4088,
+      "step": 2055
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.66841530788254e-07,
+      "loss": 0.3763,
+      "step": 2056
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.61015825310554e-07,
+      "loss": 0.4269,
+      "step": 2057
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.55206838257219e-07,
+      "loss": 0.3441,
+      "step": 2058
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.494145803724941e-07,
+      "loss": 0.4129,
+      "step": 2059
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.436390623696911e-07,
+      "loss": 0.3885,
+      "step": 2060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.378802949311583e-07,
+      "loss": 0.396,
+      "step": 2061
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.321382887082564e-07,
+      "loss": 0.4025,
+      "step": 2062
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.264130543213512e-07,
+      "loss": 0.4243,
+      "step": 2063
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.207046023597866e-07,
+      "loss": 0.3841,
+      "step": 2064
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.1501294338186e-07,
+      "loss": 0.4096,
+      "step": 2065
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.093380879148161e-07,
+      "loss": 0.3899,
+      "step": 2066
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.3839,
+      "step": 2067
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.98038829466924e-07,
+      "loss": 0.3671,
+      "step": 2068
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.924144473850827e-07,
+      "loss": 0.388,
+      "step": 2069
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.868069106121002e-07,
+      "loss": 0.3901,
+      "step": 2070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.812162295196247e-07,
+      "loss": 0.3819,
+      "step": 2071
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.756424144481313e-07,
+      "loss": 0.4075,
+      "step": 2072
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.700854757068988e-07,
+      "loss": 0.3965,
+      "step": 2073
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.3771,
+      "step": 2074
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.590222682962402e-07,
+      "loss": 0.3653,
+      "step": 2075
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.535160200892234e-07,
+      "loss": 0.4129,
+      "step": 2076
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.480266891372469e-07,
+      "loss": 0.415,
+      "step": 2077
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.425542855933299e-07,
+      "loss": 0.4111,
+      "step": 2078
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.370988195791807e-07,
+      "loss": 0.3625,
+      "step": 2079
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.316603011851809e-07,
+      "loss": 0.4123,
+      "step": 2080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.262387404703654e-07,
+      "loss": 0.4558,
+      "step": 2081
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.208341474624071e-07,
+      "loss": 0.3841,
+      "step": 2082
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.154465321575889e-07,
+      "loss": 0.3999,
+      "step": 2083
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.100759045207962e-07,
+      "loss": 0.421,
+      "step": 2084
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.047222744854943e-07,
+      "loss": 0.4063,
+      "step": 2085
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.993856519537102e-07,
+      "loss": 0.3878,
+      "step": 2086
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.940660467960137e-07,
+      "loss": 0.3711,
+      "step": 2087
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.887634688515e-07,
+      "loss": 0.4097,
+      "step": 2088
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.83477927927766e-07,
+      "loss": 0.3727,
+      "step": 2089
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.78209433800905e-07,
+      "loss": 0.4284,
+      "step": 2090
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.729579962154743e-07,
+      "loss": 0.3639,
+      "step": 2091
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.677236248844855e-07,
+      "loss": 0.4161,
+      "step": 2092
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.625063294893864e-07,
+      "loss": 0.4417,
+      "step": 2093
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.573061196800414e-07,
+      "loss": 0.3603,
+      "step": 2094
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.521230050747086e-07,
+      "loss": 0.3603,
+      "step": 2095
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.46956995260033e-07,
+      "loss": 0.369,
+      "step": 2096
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.418080997910204e-07,
+      "loss": 0.41,
+      "step": 2097
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.366763281910195e-07,
+      "loss": 0.4103,
+      "step": 2098
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.315616899517098e-07,
+      "loss": 0.3751,
+      "step": 2099
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.26464194533083e-07,
+      "loss": 0.3959,
+      "step": 2100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.213838513634186e-07,
+      "loss": 0.375,
+      "step": 2101
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.163206698392744e-07,
+      "loss": 0.3843,
+      "step": 2102
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.11274659325465e-07,
+      "loss": 0.4088,
+      "step": 2103
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.0624582915505e-07,
+      "loss": 0.4603,
+      "step": 2104
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.01234188629304e-07,
+      "loss": 0.4374,
+      "step": 2105
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.962397470177162e-07,
+      "loss": 0.4182,
+      "step": 2106
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.912625135579587e-07,
+      "loss": 0.4062,
+      "step": 2107
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.863024974558785e-07,
+      "loss": 0.3881,
+      "step": 2108
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.813597078854772e-07,
+      "loss": 0.3796,
+      "step": 2109
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.764341539888941e-07,
+      "loss": 0.3761,
+      "step": 2110
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.715258448763928e-07,
+      "loss": 0.4243,
+      "step": 2111
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.666347896263326e-07,
+      "loss": 0.3742,
+      "step": 2112
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.617609972851702e-07,
+      "loss": 0.3596,
+      "step": 2113
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.569044768674249e-07,
+      "loss": 0.4294,
+      "step": 2114
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.520652373556746e-07,
+      "loss": 0.3463,
+      "step": 2115
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.472432877005341e-07,
+      "loss": 0.3985,
+      "step": 2116
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.424386368206392e-07,
+      "loss": 0.4149,
+      "step": 2117
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.37651293602628e-07,
+      "loss": 0.4097,
+      "step": 2118
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.328812669011275e-07,
+      "loss": 0.3991,
+      "step": 2119
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.281285655387381e-07,
+      "loss": 0.3939,
+      "step": 2120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.233931983060104e-07,
+      "loss": 0.3809,
+      "step": 2121
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.186751739614405e-07,
+      "loss": 0.3842,
+      "step": 2122
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.139745012314424e-07,
+      "loss": 0.3714,
+      "step": 2123
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.092911888103404e-07,
+      "loss": 0.3809,
+      "step": 2124
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.046252453603452e-07,
+      "loss": 0.4289,
+      "step": 2125
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.999766795115447e-07,
+      "loss": 0.3781,
+      "step": 2126
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.953454998618857e-07,
+      "loss": 0.4304,
+      "step": 2127
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.907317149771552e-07,
+      "loss": 0.4003,
+      "step": 2128
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.861353333909692e-07,
+      "loss": 0.3667,
+      "step": 2129
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.815563636047539e-07,
+      "loss": 0.3689,
+      "step": 2130
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.769948140877313e-07,
+      "loss": 0.3989,
+      "step": 2131
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.724506932769014e-07,
+      "loss": 0.36,
+      "step": 2132
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6792400957703e-07,
+      "loss": 0.3668,
+      "step": 2133
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.634147713606342e-07,
+      "loss": 0.4211,
+      "step": 2134
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.589229869679558e-07,
+      "loss": 0.4569,
+      "step": 2135
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.544486647069614e-07,
+      "loss": 0.4048,
+      "step": 2136
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.499918128533155e-07,
+      "loss": 0.3838,
+      "step": 2137
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.455524396503742e-07,
+      "loss": 0.4326,
+      "step": 2138
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.411305533091604e-07,
+      "loss": 0.3902,
+      "step": 2139
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.367261620083575e-07,
+      "loss": 0.4258,
+      "step": 2140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.323392738942879e-07,
+      "loss": 0.4195,
+      "step": 2141
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.279698970809011e-07,
+      "loss": 0.4051,
+      "step": 2142
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.236180396497536e-07,
+      "loss": 0.3994,
+      "step": 2143
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.192837096500058e-07,
+      "loss": 0.3678,
+      "step": 2144
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.149669150983938e-07,
+      "loss": 0.4023,
+      "step": 2145
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.10667663979224e-07,
+      "loss": 0.4211,
+      "step": 2146
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.063859642443536e-07,
+      "loss": 0.3433,
+      "step": 2147
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.02121823813172e-07,
+      "loss": 0.4169,
+      "step": 2148
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.978752505726003e-07,
+      "loss": 0.3773,
+      "step": 2149
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.93646252377058e-07,
+      "loss": 0.3759,
+      "step": 2150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.4137,
+      "step": 2151
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.852410123762164e-07,
+      "loss": 0.3727,
+      "step": 2152
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.810647861171736e-07,
+      "loss": 0.4276,
+      "step": 2153
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.769061659956464e-07,
+      "loss": 0.4337,
+      "step": 2154
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7276515970338223e-07,
+      "loss": 0.4189,
+      "step": 2155
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6864177489954975e-07,
+      "loss": 0.419,
+      "step": 2156
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.64536019210724e-07,
+      "loss": 0.4342,
+      "step": 2157
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6044790023087373e-07,
+      "loss": 0.3915,
+      "step": 2158
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5637742552134824e-07,
+      "loss": 0.3965,
+      "step": 2159
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5232460261085966e-07,
+      "loss": 0.3872,
+      "step": 2160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4828943899547285e-07,
+      "loss": 0.3656,
+      "step": 2161
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 0.3855,
+      "step": 2162
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.402721194709436e-07,
+      "loss": 0.385,
+      "step": 2163
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3628997839056385e-07,
+      "loss": 0.4038,
+      "step": 2164
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.323255262627846e-07,
+      "loss": 0.4019,
+      "step": 2165
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2837877042021915e-07,
+      "loss": 0.3907,
+      "step": 2166
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2444971816275495e-07,
+      "loss": 0.3992,
+      "step": 2167
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2053837675752996e-07,
+      "loss": 0.4118,
+      "step": 2168
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1664475343892727e-07,
+      "loss": 0.3972,
+      "step": 2169
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.127688554085574e-07,
+      "loss": 0.3811,
+      "step": 2170
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0891068983524795e-07,
+      "loss": 0.3886,
+      "step": 2171
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.3931,
+      "step": 2172
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.012475845711106e-07,
+      "loss": 0.4165,
+      "step": 2173
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.974426590538927e-07,
+      "loss": 0.4058,
+      "step": 2174
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9365549434092987e-07,
+      "loss": 0.4126,
+      "step": 2175
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8988609743692896e-07,
+      "loss": 0.3992,
+      "step": 2176
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8613447531373196e-07,
+      "loss": 0.396,
+      "step": 2177
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.82400634910306e-07,
+      "loss": 0.3972,
+      "step": 2178
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7868458313272906e-07,
+      "loss": 0.43,
+      "step": 2179
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7498632685417646e-07,
+      "loss": 0.3854,
+      "step": 2180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7130587291490995e-07,
+      "loss": 0.3859,
+      "step": 2181
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6764322812226416e-07,
+      "loss": 0.4181,
+      "step": 2182
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6399839925063683e-07,
+      "loss": 0.3752,
+      "step": 2183
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.603713930414676e-07,
+      "loss": 0.345,
+      "step": 2184
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5676221620323583e-07,
+      "loss": 0.3956,
+      "step": 2185
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.531708754114438e-07,
+      "loss": 0.3874,
+      "step": 2186
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4959737730860144e-07,
+      "loss": 0.446,
+      "step": 2187
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4604172850421946e-07,
+      "loss": 0.3936,
+      "step": 2188
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.425039355747961e-07,
+      "loss": 0.4547,
+      "step": 2189
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3898400506379937e-07,
+      "loss": 0.3894,
+      "step": 2190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.354819434816614e-07,
+      "loss": 0.4004,
+      "step": 2191
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.319977573057642e-07,
+      "loss": 0.4224,
+      "step": 2192
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2853145298042954e-07,
+      "loss": 0.3453,
+      "step": 2193
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.250830369168989e-07,
+      "loss": 0.3827,
+      "step": 2194
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2165251549333585e-07,
+      "loss": 0.4247,
+      "step": 2195
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1823989505479934e-07,
+      "loss": 0.4058,
+      "step": 2196
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1484518191324253e-07,
+      "loss": 0.3817,
+      "step": 2197
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.114683823474962e-07,
+      "loss": 0.3922,
+      "step": 2198
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.081095026032599e-07,
+      "loss": 0.4182,
+      "step": 2199
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0476854889308737e-07,
+      "loss": 0.4578,
+      "step": 2200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0144552739637565e-07,
+      "loss": 0.3712,
+      "step": 2201
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9814044425935605e-07,
+      "loss": 0.4253,
+      "step": 2202
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9485330559507976e-07,
+      "loss": 0.4499,
+      "step": 2203
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.915841174834111e-07,
+      "loss": 0.379,
+      "step": 2204
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8833288597100996e-07,
+      "loss": 0.4167,
+      "step": 2205
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8509961707132496e-07,
+      "loss": 0.4048,
+      "step": 2206
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.818843167645835e-07,
+      "loss": 0.3951,
+      "step": 2207
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7868699099777295e-07,
+      "loss": 0.3864,
+      "step": 2208
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7550764568464174e-07,
+      "loss": 0.3984,
+      "step": 2209
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7234628670567585e-07,
+      "loss": 0.3821,
+      "step": 2210
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.69202919908097e-07,
+      "loss": 0.3893,
+      "step": 2211
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6607755110584886e-07,
+      "loss": 0.4154,
+      "step": 2212
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6297018607958745e-07,
+      "loss": 0.4245,
+      "step": 2213
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5988083057666534e-07,
+      "loss": 0.4286,
+      "step": 2214
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5680949031112625e-07,
+      "loss": 0.4113,
+      "step": 2215
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.53756170963696e-07,
+      "loss": 0.3748,
+      "step": 2216
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.4114,
+      "step": 2217
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.477036175793823e-07,
+      "loss": 0.3748,
+      "step": 2218
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.447043947372507e-07,
+      "loss": 0.4057,
+      "step": 2219
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.417232152027016e-07,
+      "loss": 0.3702,
+      "step": 2220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.387600844896998e-07,
+      "loss": 0.4005,
+      "step": 2221
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3581500807882462e-07,
+      "loss": 0.3972,
+      "step": 2222
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3288799141726548e-07,
+      "loss": 0.4112,
+      "step": 2223
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2997903991880288e-07,
+      "loss": 0.3751,
+      "step": 2224
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2708815896381075e-07,
+      "loss": 0.428,
+      "step": 2225
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.242153538992331e-07,
+      "loss": 0.383,
+      "step": 2226
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2136063003858733e-07,
+      "loss": 0.4111,
+      "step": 2227
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.408,
+      "step": 2228
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1570544701592254e-07,
+      "loss": 0.3938,
+      "step": 2229
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1290499831367994e-07,
+      "loss": 0.3643,
+      "step": 2230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1012265173490197e-07,
+      "loss": 0.3946,
+      "step": 2231
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0735841242578992e-07,
+      "loss": 0.3946,
+      "step": 2232
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.046122854990562e-07,
+      "loss": 0.4248,
+      "step": 2233
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.018842760339146e-07,
+      "loss": 0.4896,
+      "step": 2234
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9917438907606556e-07,
+      "loss": 0.3924,
+      "step": 2235
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9648262963769315e-07,
+      "loss": 0.4199,
+      "step": 2236
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.938090026974504e-07,
+      "loss": 0.3618,
+      "step": 2237
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9115351320045495e-07,
+      "loss": 0.4114,
+      "step": 2238
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.885161660582746e-07,
+      "loss": 0.4296,
+      "step": 2239
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8589696614892406e-07,
+      "loss": 0.3613,
+      "step": 2240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8329591831685146e-07,
+      "loss": 0.4157,
+      "step": 2241
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8071302737293294e-07,
+      "loss": 0.3904,
+      "step": 2242
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7814829809446044e-07,
+      "loss": 0.4846,
+      "step": 2243
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7560173522513268e-07,
+      "loss": 0.3875,
+      "step": 2244
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.730733434750509e-07,
+      "loss": 0.4001,
+      "step": 2245
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7056312752070646e-07,
+      "loss": 0.4412,
+      "step": 2246
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6807109200496995e-07,
+      "loss": 0.3614,
+      "step": 2247
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.65597241537091e-07,
+      "loss": 0.3812,
+      "step": 2248
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.631415806926795e-07,
+      "loss": 0.4462,
+      "step": 2249
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6070411401370335e-07,
+      "loss": 0.4275,
+      "step": 2250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5828484600848072e-07,
+      "loss": 0.342,
+      "step": 2251
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.558837811516667e-07,
+      "loss": 0.3814,
+      "step": 2252
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5350092388425108e-07,
+      "loss": 0.4111,
+      "step": 2253
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5113627861354396e-07,
+      "loss": 0.4014,
+      "step": 2254
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4878984971317234e-07,
+      "loss": 0.3861,
+      "step": 2255
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.464616415230702e-07,
+      "loss": 0.3839,
+      "step": 2256
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4415165834947064e-07,
+      "loss": 0.4371,
+      "step": 2257
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4185990446489718e-07,
+      "loss": 0.3824,
+      "step": 2258
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3958638410815905e-07,
+      "loss": 0.3943,
+      "step": 2259
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3733110148433926e-07,
+      "loss": 0.3761,
+      "step": 2260
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.350940607647866e-07,
+      "loss": 0.3481,
+      "step": 2261
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3287526608711132e-07,
+      "loss": 0.4089,
+      "step": 2262
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3067472155517736e-07,
+      "loss": 0.3452,
+      "step": 2263
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2849243123909227e-07,
+      "loss": 0.4193,
+      "step": 2264
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2632839917520179e-07,
+      "loss": 0.4041,
+      "step": 2265
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.241826293660786e-07,
+      "loss": 0.3523,
+      "step": 2266
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2205512578051915e-07,
+      "loss": 0.3897,
+      "step": 2267
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1994589235353682e-07,
+      "loss": 0.4425,
+      "step": 2268
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1785493298634875e-07,
+      "loss": 0.3883,
+      "step": 2269
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1578225154637579e-07,
+      "loss": 0.3717,
+      "step": 2270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1372785186723135e-07,
+      "loss": 0.4065,
+      "step": 2271
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 0.4039,
+      "step": 2272
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0967391295680363e-07,
+      "loss": 0.4244,
+      "step": 2273
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0767438122364914e-07,
+      "loss": 0.4162,
+      "step": 2274
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0569314624756855e-07,
+      "loss": 0.3131,
+      "step": 2275
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0373021169303387e-07,
+      "loss": 0.3559,
+      "step": 2276
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.3504,
+      "step": 2277
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.985925833725706e-08,
+      "loss": 0.4048,
+      "step": 2278
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.795124669569445e-08,
+      "loss": 0.388,
+      "step": 2279
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.606154979502791e-08,
+      "loss": 0.3921,
+      "step": 2280
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.419017113042273e-08,
+      "loss": 0.3907,
+      "step": 2281
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.233711416316571e-08,
+      "loss": 0.4239,
+      "step": 2282
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.0502382320653e-08,
+      "loss": 0.3773,
+      "step": 2283
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.868597899638897e-08,
+      "loss": 0.3959,
+      "step": 2284
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.688790754997623e-08,
+      "loss": 0.3866,
+      "step": 2285
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.510817130711224e-08,
+      "loss": 0.3805,
+      "step": 2286
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.334677355958055e-08,
+      "loss": 0.4308,
+      "step": 2287
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.160371756524843e-08,
+      "loss": 0.4182,
+      "step": 2288
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.987900654805481e-08,
+      "loss": 0.4131,
+      "step": 2289
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.817264369801125e-08,
+      "loss": 0.3999,
+      "step": 2290
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.648463217118985e-08,
+      "loss": 0.4211,
+      "step": 2291
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.481497508972313e-08,
+      "loss": 0.4134,
+      "step": 2292
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.316367554179194e-08,
+      "loss": 0.4014,
+      "step": 2293
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.153073658162646e-08,
+      "loss": 0.3517,
+      "step": 2294
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.991616122949629e-08,
+      "loss": 0.384,
+      "step": 2295
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.831995247170597e-08,
+      "loss": 0.3784,
+      "step": 2296
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.674211326058721e-08,
+      "loss": 0.3978,
+      "step": 2297
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.51826465144978e-08,
+      "loss": 0.4126,
+      "step": 2298
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.36415551178149e-08,
+      "loss": 0.4171,
+      "step": 2299
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.21188419209251e-08,
+      "loss": 0.4217,
+      "step": 2300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.061450974022775e-08,
+      "loss": 0.3884,
+      "step": 2301
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.912856135812051e-08,
+      "loss": 0.394,
+      "step": 2302
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.766099952300041e-08,
+      "loss": 0.3639,
+      "step": 2303
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.621182694925731e-08,
+      "loss": 0.395,
+      "step": 2304
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 0.4383,
+      "step": 2305
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.33686602733885e-08,
+      "loss": 0.4066,
+      "step": 2306
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.19746714299596e-08,
+      "loss": 0.4317,
+      "step": 2307
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.05990823652891e-08,
+      "loss": 0.4152,
+      "step": 2308
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9241895623654e-08,
+      "loss": 0.4356,
+      "step": 2309
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.79031137152941e-08,
+      "loss": 0.4044,
+      "step": 2310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.658273911640976e-08,
+      "loss": 0.3967,
+      "step": 2311
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.528077426915412e-08,
+      "loss": 0.3433,
+      "step": 2312
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.399722158162867e-08,
+      "loss": 0.38,
+      "step": 2313
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.273208342788104e-08,
+      "loss": 0.4221,
+      "step": 2314
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1485362147900555e-08,
+      "loss": 0.3986,
+      "step": 2315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.025706004760932e-08,
+      "loss": 0.3958,
+      "step": 2316
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.904717939886671e-08,
+      "loss": 0.4569,
+      "step": 2317
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.785572243945601e-08,
+      "loss": 0.4349,
+      "step": 2318
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.668269137308666e-08,
+      "loss": 0.3857,
+      "step": 2319
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.552808836938426e-08,
+      "loss": 0.3771,
+      "step": 2320
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4391915563892766e-08,
+      "loss": 0.4013,
+      "step": 2321
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.327417505806785e-08,
+      "loss": 0.4313,
+      "step": 2322
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.217486891927135e-08,
+      "loss": 0.3918,
+      "step": 2323
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1093999180772385e-08,
+      "loss": 0.3776,
+      "step": 2324
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.003156784173511e-08,
+      "loss": 0.3848,
+      "step": 2325
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.898757686722542e-08,
+      "loss": 0.3876,
+      "step": 2326
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.796202818819871e-08,
+      "loss": 0.3754,
+      "step": 2327
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.695492370149988e-08,
+      "loss": 0.4024,
+      "step": 2328
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5966265269862233e-08,
+      "loss": 0.3886,
+      "step": 2329
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.499605472189748e-08,
+      "loss": 0.3885,
+      "step": 2330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4044293852099055e-08,
+      "loss": 0.4255,
+      "step": 2331
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.311098442083659e-08,
+      "loss": 0.3908,
+      "step": 2332
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.219612815434924e-08,
+      "loss": 0.4252,
+      "step": 2333
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1299726744747896e-08,
+      "loss": 0.3531,
+      "step": 2334
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0421781850008536e-08,
+      "loss": 0.4255,
+      "step": 2335
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.956229509397001e-08,
+      "loss": 0.3899,
+      "step": 2336
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.872126806633068e-08,
+      "loss": 0.4122,
+      "step": 2337
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7898702322648453e-08,
+      "loss": 0.3939,
+      "step": 2338
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.709459938433189e-08,
+      "loss": 0.3616,
+      "step": 2339
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.630896073864352e-08,
+      "loss": 0.4058,
+      "step": 2340
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.554178783869209e-08,
+      "loss": 0.4602,
+      "step": 2341
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4793082103435885e-08,
+      "loss": 0.4123,
+      "step": 2342
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4062844917672735e-08,
+      "loss": 0.3742,
+      "step": 2343
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3351077632044462e-08,
+      "loss": 0.445,
+      "step": 2344
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2657781563026883e-08,
+      "loss": 0.4254,
+      "step": 2345
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1982957992936472e-08,
+      "loss": 0.3782,
+      "step": 2346
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1326608169920373e-08,
+      "loss": 0.3985,
+      "step": 2347
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0688733307956389e-08,
+      "loss": 0.3899,
+      "step": 2348
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0069334586854106e-08,
+      "loss": 0.3529,
+      "step": 2349
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.468413152244893e-09,
+      "loss": 0.3663,
+      "step": 2350
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.885970115590781e-09,
+      "loss": 0.3897,
+      "step": 2351
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.322006554171147e-09,
+      "loss": 0.3682,
+      "step": 2352
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.776523511089373e-09,
+      "loss": 0.4183,
+      "step": 2353
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.2495219952639636e-09,
+      "loss": 0.3771,
+      "step": 2354
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.741002981435207e-09,
+      "loss": 0.3695,
+      "step": 2355
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.250967410152964e-09,
+      "loss": 0.3526,
+      "step": 2356
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.779416187783327e-09,
+      "loss": 0.3461,
+      "step": 2357
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.326350186503071e-09,
+      "loss": 0.4118,
+      "step": 2358
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.891770244298544e-09,
+      "loss": 0.3868,
+      "step": 2359
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.475677164966774e-09,
+      "loss": 0.3743,
+      "step": 2360
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0780717181077015e-09,
+      "loss": 0.374,
+      "step": 2361
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.698954639129726e-09,
+      "loss": 0.3711,
+      "step": 2362
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3383266292452697e-09,
+      "loss": 0.4243,
+      "step": 2363
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9961883554674443e-09,
+      "loss": 0.4208,
+      "step": 2364
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.672540450612271e-09,
+      "loss": 0.3658,
+      "step": 2365
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.367383513296462e-09,
+      "loss": 0.4194,
+      "step": 2366
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.080718107935198e-09,
+      "loss": 0.3667,
+      "step": 2367
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8125447647421302e-09,
+      "loss": 0.3662,
+      "step": 2368
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5628639797293788e-09,
+      "loss": 0.4073,
+      "step": 2369
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3316762147030925e-09,
+      "loss": 0.4037,
+      "step": 2370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1189818972656697e-09,
+      "loss": 0.3955,
+      "step": 2371
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.24781420816867e-10,
+      "loss": 0.4268,
+      "step": 2372
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.490751445449196e-10,
+      "loss": 0.4304,
+      "step": 2373
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.918633934376416e-10,
+      "loss": 0.4177,
+      "step": 2374
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.531464582713252e-10,
+      "loss": 0.3867,
+      "step": 2375
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.3292459561518055e-10,
+      "loss": 0.3754,
+      "step": 2376
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3119802783022615e-10,
+      "loss": 0.4391,
+      "step": 2377
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4796694307039894e-10,
+      "loss": 0.4269,
+      "step": 2378
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.323149527811325e-11,
+      "loss": 0.4384,
+      "step": 2379
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.6991804187591626e-11,
+      "loss": 0.4184,
+      "step": 2380
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.24795532264433e-12,
+      "loss": 0.4226,
+      "step": 2381
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.4031,
+      "step": 2382
+    },
+    {
+      "epoch": 3.0,
+      "step": 2382,
+      "total_flos": 1.1071453981258547e+18,
+      "train_loss": 0.6899283054893903,
+      "train_runtime": 12615.1361,
+      "train_samples_per_second": 24.167,
+      "train_steps_per_second": 0.189
+    }
+  ],
+  "max_steps": 2382,
+  "num_train_epochs": 3,
+  "total_flos": 1.1071453981258547e+18,
+  "trial_name": null,
+  "trial_params": null
+}